Method for locating a mobile node assisted by an artificial intelligence model and a visibility map

Abstract

Method for locating a mobile node comprising: Receiving N≥ 4 radio frequency measurements corresponding to signals exchanged between a mobile node to be located and N reference nodes; Receiving a 3D map of said area; Determining a first set of residuals (MR) calculated for several subsets of measurements, each comprising at most N-1 measurements; For a plurality of given positions: Determining (902) a second set (ML) of indicators of the radio propagation paths between said position and that of each reference node; Executing (903) an inference phase of an artificial intelligence model; Determining (904) a provisional position of the mobile node; Calculating (905) a consistency score between the provisional position and the given position, and determining the final position of the mobile node as a function of the consistency score. Figure for the abstract: Fig. 2

Description

Title of the invention: Method for locating a mobile node assisted by an artificial intelligence model and a visibility map

[0001] The invention relates to methods and systems for locating a mobile node in an environment by means of radio frequency measurements performed on signals exchanged between this mobile node and several reference nodes. The measurements relate to the time of flight of the exchanged signals or to distances calculated from the time of flight. The reference nodes may be terrestrial communication stations, in which case the application area is that of radio localization. The reference nodes may also be satellites, in which case the application area is that of satellite radionavigation.

[0002] The invention relates more specifically to a method and a localization system (radio or satellite) assisted by an artificial intelligence model and a visibility map of the environment of the mobile node and the reference nodes.

[0003] In the field of satellite positioning, pseudo-range or phase measurements can be used to determine the position as well as the clock offset of the receiver, while Doppler offset measurement is used for determining the speed and clock drift of the receiver.

[0004] For example, if we consider a pseudo-distance measurement Pj between a receiver with coordinates x' y* z and a satellite SVi with coordinates xr y? zi, this measurement respects the following relationship:

[0005] - A++ (1) ll\ / *

[0006] Where is the geometric distance between the receiver and the SVi satellite,

[0007] d^x, y, zj =

[0008] 5 is a distance offset resulting from a clock offset A t between the receiver and the constellation (all satellites belonging to the same GNSS constellation being perfectly synchronized, the offset is the same for all satellites in the constellation) converted into meters by multiplying it by the speed of light c

[0009] <5=A / .c(3)

[0010] £' is a measurement error caused by multiple phenomena such as: - the measurement noise of the receiver, on the order of a few centimeters to a few meters, - propagation delays in the ionosphere and troposphere, on the order of a few centimeters to a few meters, - errors in satellite positioning, - effects related to signal propagation.

[0011] The operation of the satellite radionavigation system is such that observations from different satellites are collected simultaneously during a period of time called "epoch" (epoch = simultaneous transmissions of satellite signals) at the end of which a set of measurements [is collected.

[0012] The calculation of the positional solution (x, y, z, ô) depends on the operating mode of the receiver. A distinction is made between a tracking mode and a so-called “single-epoch” mode.

[0013] The tracking mode is a receiver operating mode for which a solution has already been calculated at a previous epoch k-1. In this case, the solution at time k is obtained by updating the solution k-1 using the measurements collected at epoch k. This calculation is generally performed by a Kalman filter or one of its many derivatives (EKF, UKF, etc.).

[0014] The so-called "single-epoch" mode is a receiver operating mode for which no prior solution exists. This mode, while a priori less efficient than the tracking mode, is nevertheless useful for certain applications such as: - the initialization of the tracking algorithm, its use in a fusion context with other measurement modalities such as inertial measurement units, where the calculated solutions are filtered again by another algorithm, in certain applications (e.g., IoT for Internet of Things) where the calculation is done remotely (in a remote server) and for which various constraints (consumption, throughput) only allow a one-off transfer of measurements from the sensor nodes / tags to be located.

[0015] The invention relates more particularly, but not exclusively, to the so-called "single-epoch" mode in that it aims at calculating a positioning solution for a given epoch at time k without benefiting from a previous solution.

[0016] In so-called single-epoch mode, the calculation of a position solution (X = [x, y, z, 5]) comprising four unknowns can therefore be determined from the set of measurements { / x} collected at epoch k using equation (1). Since the number of measurements N (typically, from 4 to 50) is generally greater than the number of unknowns, which is at least 4, the solution must therefore be calculated according to a criterion for minimizing the errors affecting the measurements—also called residuals—such as least squares. More precisely, the number of unknowns is equal to 4 when all the received signals are emitted by satellites of the same constellation (for example, the GPS constellation). If the signals are emitted by satellites from at least two constellations (e.g., GPS and GALILEO) then the number of unknowns is equal to 5. In general, the number of unknowns is equal to 3+m, where m is the number of satellite constellations considered.

[0017] Relation (4) gives the least squares equation allowing the solution of position X to be determined by searching for the solution which minimizes a weighted sum of the squared residuals.

[0018] (x,y,z,ô) = argmin^L. A / x)

[0019] The residuals .(x) correspond to the difference between the prediction of the measure according to the model of equation (1) for the solution X on the one hand, and the measurement Pi carried out on the other hand, where x,y,z and ô are the solutions of equation (4).

[0020] _ p. _ + (5)

[0021] And <oi définit le coefficient de pondération associé à la mesure #i, c’est-à-dire « l’importance » que celle-ci aura dans le calcul de la solution.

[0022] The term defines the weighting coefficient or weight associated with each measure Pi. The higher this coefficient, the greater the contribution of the associated measure to the calculated solution. A weight of zero (07 = 0) means that a measure has no influence on the calculated solution and is therefore equivalent to the absence of that measure. If the weight is non-zero, its importance is evaluated relative to the weights of the other measures; its value alone does not define its importance.

[0023] There are several weighting strategies for solving the least squares given by equation (4). One possible solution is to calculate the weight as a function of the probable error associated with each measurement using the model of equation (6), i.e. the inverse of the variance <7? of the probable error on the measurement of index i.

[0024] W.= A(6)

[0025] This weighting strategy does indeed have theoretical foundations (Bayesian estimation), which allow it to be justified under certain assumptions. In this case, the probable error is the standard deviation associated with the measurement error distribution P, that is, the distribution of the random term e' of equation (1). In addition to its theoretical justification, this approach makes it possible to link the value expressed in meters to measurable physical phenomena. Thus, the distribution of measurement errors can be established empirically through test campaigns, or from other known elements (satellite positioning accuracy, propagation in the ionosphere, receiver noise, etc.). Typically, the values ​​a> are between 0.1 m and 20 m, but can vary depending on the intensity of the received signal or the satellite's elevation, for example.

[0026] Techniques linking different ancillary metrics (signal-to-noise ratio C / N0, satellite elevation, etc.) to the assumed accuracy of the measurement are called stochastic weighting.

[0027] For example, the variance of the error can be obtained by means of an empirical model given by equation (7).

[0028] 2__l_ / .,^(7)

[0029] In which and V; represent respectively the elevation angle and the intensity of the signal received from the SVi satellite, and crf and are two empirical terms.

[0030] Other empirical models may be used in place of the one given above.

[0031] While these empirical formulas are widely used, they remain difficult to calibrate, and a more refined modeling of the distribution of each of the received measurements would improve performance. However, this more refined modeling can prove extremely complex in practice, given the large number of phenomena involved (state of atmospheric layers, type of receiver, satellite positioning accuracy, presence of obstacles in the receiver's environment, etc.) and the difficulty in predicting them.

[0032] Furthermore, in order to guarantee the optimality, and therefore good accuracy, of the calculated solutions, it is important that all the measures taken into account to establish this solution conform to the model used a priori, that is to say within a range of expected error.

[0033] In the field of satellite navigation systems, there is therefore a need for a localization method which makes it possible to better take into account the uncertainties linked to the theoretical models of calculation of the weighting coefficients when solving the solution by least squares.

[0034] The same problems exist in the field of radio localization.

[0035] In particular, equations (1) to (5) also apply to signals exchanged between a mobile node and ground stations. In this case, the coordinates xr > Vf zi are those of the ground stations and not of the satellites. The same problems of clock differences and temporal synchronization hazard exist. The residuals can be calculated as differences in time-of-flight or distance measurements.

[0036] Moreover, in both fields of application, existing localization solutions do not always take into account the impact of obstacles, particularly in urban environments, and in particular the influence of reflections or diffractions of signals on these obstacles.

[0037] Depending on the environment in which the wave propagates, it may be reflected (this is called multipath propagation) and / or obstructed. The term Line Of Sight (LOS) is used when the wave propagates along a direct path, without reflection, between the transmitter and the receiver, and Non-Line Of Sight (NLOS) is used in the opposite case.

[0038] This occurs for example in urban environments where buildings can act as both obstacles and / or reflectors for electromagnetic fields.

[0039] Fig. 1 illustrates these different situations.

[0040] The reference node A1 is in a LOS situation without multipath with respect to the moving node TM because it only receives the direct path, without a reflected path. The reference node A2 is in a LOS situation with multipath with respect to the moving node TM because it receives the direct path A1 as well as a reflected path. The reference node A3 is in a NLOS situation with multipath with respect to the moving node TM because it does not receive the direct path Isa, which is obstructed by building B1, but it does receive a reflected path Lb by building B2 as well as a diffracted path by building B1.

[0041] An NLOS situation without multipath leads to the non-reception of the signal since it has no path to reach its destination.

[0042] In the representation in Figure 1, the measurement performed by the reference node Al can be considered the most reliable because it is not subject to any disturbance in its propagation. The associated error (Σi) is in this case mainly determined by the measurement noise, which constitutes a minimal error.

[0043] For example, for a LoRa type waveform, the error caused by noise alone is typically on the order of a hundred meters (between 10 m and 300 m).

[0044] The measurement performed by the fixed terminal A2 can be disrupted by the presence of multipath propagation. The resulting error depends on the ability of the reference node A2 to distinguish the direct path from the reflected path (which will arrive later than the direct path), as well as on the relative power of the latter compared to the direct path. This is therefore an intermediate situation.

[0045] For example, for a LoRa waveform, the error caused by noise and multipath propagation varies greatly depending on the characteristics of the reflected paths. The errors can range from a few tens of meters (identical to noise alone) to several kilometers.

[0046] The measurement performed by the reference node A3 is, however, highly biased. Indeed, the only signal reaching this reference node has traveled a distance corresponding to path ¢33, significantly greater than the direct path t^A; the time of flight therefore reflects this distance and not that of the direct path. This measurement is generally considered an artifact (or outlier) because it exhibits an error that can be several orders of magnitude greater than the error due to noise alone.

[0047] For example, for a LoRa type radio, the error in an NLOS situation is typically from several hundred meters to several kilometers.

[0048] The difficulty of such a situation is that it is very difficult, if not impossible, to determine the reliability of each of the measures.

[0049] Another technical problem to be solved in the field of localization in environments including obstacles is to take into account the impact of these obstacles on the quality of the signals and therefore of the measurements taken to improve the accuracy of the final localization.

[0050] The Applicant's French patent application published under number FR3136065 describes a method for locating a mobile node by radio measurements using a visibility map.

[0051] The method proposed in this application uses information about the environment of the mobile node and the reference nodes in the form of an environment map, which allows for the creation of a visibility map for different hypothetical positions of the mobile node within the map. The node's location is determined by calculating a consistency score for each position hypothesis, the consistency score being based on a visibility indicator between the assumed position of the mobile node and the reference node.

[0052] The Applicant's French patent application filed under number FR2304055 relates to a localization method using satellite radionavigation signals which uses an artificial intelligence model to determine a set of optimal weighting coefficients used to determine the position of the moving node by means of a least squares resolution applied to a set of residuals weighted by these coefficients.

[0053] The two aforementioned solutions do not fully take into account mapping information to accurately estimate the impact of reflected paths and diffracted paths on obstacles in calculating the position of the moving node.

[0054] The invention relates to a new radio or satellite localization method that uses additional information concerning the signal paths between the mobile node and the reference nodes from a 3D map of the environment. This information relates to characteristics of the direct, diffracted, and reflected paths of the signal. This additional information enhances the learning of an artificial intelligence model trained to determine a position by least-squares resolution and is combined with the calculation of a consistency score determined for a plurality of positions on the map.

[0055] An object of the invention is a computer-implemented method for training an artificial intelligence model intended to be used to determine the position of a moving node in a given area, the method comprising the steps of: - Receive a training dataset comprising several sets of radio frequency measurements corresponding to signals exchanged respectively between a mobile node of known position and several reference nodes of known positions, and a 3D map representing the environment of the mobile node and the reference nodes, - Determine, for each set of measurements, a set of metrics comprising at least a first set of residuals calculated for several subsets of measurements, each excluding at least one measurement from the set, and a second set of indicators of radio propagation paths between the mobile node and each reference node, taking into account environmental obstacles represented in the 3D map. - For each set of measurements, i. Determine a set of reference weighting coefficients, ii. Train the artificial intelligence model to produce a set of weighting coefficients from the metric sets of the training data, in such a way as to minimize the distance between said weighting coefficients and the reference weighting coefficients, - the weighting coefficients being intended to weight a set of residuals, equal to a difference between a measurement and a predicted positioning information, during a calculation of prediction of a positioning information by minimizing the sum of the residuals squared and weighted by the weighting coefficients.

[0056] According to particular embodiments of such a method:

[0057] - each reference weighting coefficient can be a function of a residual calculated from each radio frequency measurement and the distances or flight times between the moving node and each reference node.

[0058] - The artificial intelligence model can be an artificial neural network, by example a recurrent neural network.

[0059] Another object of the invention is a method for locating a moving node in a given area comprising the steps of: Receive a set of N radio frequency measurements corresponding to signals exchanged respectively between a mobile node to be located and N reference nodes of known positions in the area, N being an integer at least equal to 4, Receive a 3D map representing the said area, Determine, from the received measurements, a first set of residuals calculated for several subsets of measurements, each comprising at most Nl measurements. For a given plurality of positions on the map, i. Determine, from the 3D map, a second set indicators of radio propagation paths between said map position and the position of each reference node, taking into account environmental obstacles represented in the 3D map, Perform an inference phase of the artificial intelligence model trained using the above learning method, from the first and second sets to determine a set of weighting coefficients, iii. iv. Determine a provisional position of the moving node from of said measures and weighting coefficients by seeking the value of the position which minimizes the sum of the squared residuals, associated with the N measures, and weighted by the weighting coefficients. calculate a consistency score between the provisional position and the given position of the map determine the final position of the moving node based on the consistency score

[0060] According to particular embodiments of such a method for locating a moving node:

[0061] - The consistency score can be calculated from the sum of the high residuals squared, associated with the N measurements, and weighted by the weighting coefficients, the residuals being calculated for the provisional position.

[0062] - The final position of the moving node can be the position of the card that gives the score of highest consistency or which is equal to an average of the map positions weighted by their respective consistency scores.

[0063] According to various embodiments of the invention:

[0064] The second set of radio propagation path indicators may include at least one indicator from among a direct path indicator, a diffracted path indicator and a reflected path indicator.

[0065] - At least one direct path indicator can be taken from an indicator of the number of buildings intercepted by the path or an indicator of the total distance traveled by the signal inside the buildings crossed.

[0066] - Alternatively, at least one diffracted path indicator can be taken from a An indicator of the additional distance traveled for the diffracted path compared to the direct path, or an indicator of the diffraction angle. In this case, the method may include determining the diffracted path as the path of minimum distance that intercepts an edge of an obstacle among all the obstacles present on the 3D map.

[0067] - Alternatively, at least one reflected path indicator is taken from an indicator the additional distance traveled for the reflected path compared to the direct path, or an indicator of the reflection angle. In this case, the method may include determining the reflected path as the path of minimum distance reflected on a face of an obstacle among all the obstacles present on the 3D map.

[0068] - Radio frequency measurement can be a time-of-flight measurement and a residual is calculated as the difference between the measurement and the theoretical time of flight between the position of the moving node and the position of a reference node or the radio frequency measurement can be a distance measurement and a residual can be calculated as the difference between the measurement and the distance between the moving node and the reference node.

[0069] - Radio frequency measurements are satellite radionavigation measurements and Reference nodes can be satellites or reference nodes are ground stations.

[0070] Yet another object of the invention is a mobile node localization system comprising a processing unit configured to execute the steps of the above localization method.

[0071] Yet another object of the invention is a computer program comprising instructions for the execution of the above method, when the program is executed by a processor.

[0072] Yet another object of the invention is a processor-readable recording medium on which a program comprising instructions is recorded for the execution of the above method, when the program is executed by a processor.

[0073] Other features and advantages of the present invention will become more apparent from the following description in relation to the following accompanying drawings.

[0074] [Fig. 1] represents an example illustrating multipath situations and obstructions of signals exchanged between a mobile node and several reference nodes,

[0075] [Fig.2] represents an organizational chart detailing the general operation of the invention,

[0076] [Fig.3] represents a flowchart detailing the steps of a method learning an artificial intelligence model intended to be used to determine the location information of a mobile node, according to one embodiment of the invention,

[0077] [Fig.4] represents a first set of residues intended to power the motor artificial intelligence, as a first example of implementation,

[0078] [Fig.5] represents a second set of link characteristics intended to feed an artificial intelligence engine according to a second example of implementation.

[0079] [Fig.6] represents a diagram illustrating different paths considered to determine the second set of characteristics,

[0080] [Fig.7] represents a flowchart of a method for determining a route diffracted,

[0081] [Fig.8] represents a flowchart of a method for determining a route thoughtful,

[0082] [Fig.9] represents a flowchart detailing the steps of a method of localization of a mobile node from the artificial intelligence model trained by the method of [Fig.3], according to an embodiment of the invention,

[0083] In the following description, the invention is described in the context of a radio localization method based on measurements performed on signals transmitted according to a waveform conforming to a terrestrial radio communications system, for example, a 5G or LoRA transmission technology for LPWAN (Low Power Wide Area Network) or NB-IoT (Narrowband Internet of Things) communication networks. In this scenario, the mobile node to be located carries a radio communication device conforming to one of these communication standards, and the reference nodes are terrestrial base stations with known fixed positions that are part of the communications network infrastructure (a simple example of such a network is shown in [Fig. 1]).

[0084] The invention applies identically to satellite radionavigation systems in which the mobile node carries a satellite radionavigation device and the reference nodes are satellites whose positions are known by means of ephemerides and information transmitted in radionavigation messages.

[0085] Figure [Fig.2] schematically illustrates, on an organizational chart, the general operation of the invention.

[0086] The invention is divided into two phases. A first phase 201 during which an artificial intelligence model, for example an artificial neural network, is trained from a training dataset to learn a function f that automatically calculates the weighting coefficients œi, which are used to weight the residuals calculated from the measurements (equation 4). The learning is performed using a set of joint characteristics specific to each communication link.

[0087] In the case where the measurements are time-of-flight measurements, equation 5 for calculating a residual becomes:

[0088] A = Ti _ (8)

[0089] With hf(XV,Z) = F+tQ

[0090] (x,y,z) are the coordinates of the moving node, (x,yi,Zi) are the coordinates of the base station (reference node) and t0 is the synchronization error.

[0091] The training is performed on a number K of sets of measurements (by 11 lJ * k=LK example associated with k known positions of the moving node. The training is carried out so that the weighting coefficients obtained are as close as possible to reference coefficients { The number K of measurement sets used is such that KN is greater than the number of parameters P (KN > P) and is for example between 10000 and 10000000. N is the number of measurements taken from signals emitted by different satellites.

[0092] For example, the reference weighting coefficients are calculated using equation (6) from a known reference position solution xref which is for example provided by a high-precision receiver which may be equipped with an inertial measurement unit or any other means independent of the targeted communications network.

[0093] The parameters of the artificial intelligence model {v£} are adjusted, for example, so as to minimize a cost function C which is a function of a distance between the weighting coefficients obtained and the reference weighting coefficients.

[0094] The cost function C is given, for example, by relation (9):

[0095] XMI <9)

[0096] In a second phase 202 of the invention, the artificial intelligence model trained in the first phase 201 is then used to predict, from new measurements, a set of weighting coefficients which are used to calculate a positioning solution for a mobile node.

[0097] The location of the mobile node is achieved from measurements 204 (similar to those made to generate the training basis of the model) and a 3D map of the environment 203 of the mobile node.

[0098] The artificial intelligence model used may be an artificial neural network, for example a recurrent neural network or any other model that can be trained to learn to perform a particular function.

[0099] According to a particular embodiment of the invention, the artificial intelligence model is a recurrent neural network or LSTM-NN for "Long Short Term Memory Neural Network". The network comprises, for example, two hidden layers of 893 and 517 neurons respectively. The input layer has as many neurons as there are metrics calculated from the measurements and the 3D map of the environment, and the output layer has a number NneUr of neurons equal to the number of weighting coefficients to be generated, which is itself equal to the maximum number of measurements associated with different reference nodes that a receiver can receive. This parameter can be defined a priori.

[0100] The activation functions of the neural network layers are, for example, a hyperbolic tangent function (tanh) or a ReLU activation function. The parameters of the neural network, in particular the synaptic coefficients, are initialized to a predetermined value, for example 0 or 1 or a random value between 0 and 1.

[0101] The number of layers (1 to 100), the number of neurons (5 to 1000 per layer), and the number of parameters P (1000 to 100,000,000) can vary. The activation function for each layer can be chosen from examples such as ReLU, tanh, or sigmoid, or softmax for the output layer. The number of layers, the number of neurons, and the choice of activation function can themselves be among the parameters that can be optimized during the training step 302. Other types of networks can also be used, such as fully connected or convolutional networks. “CNN” (convolutional-NN) or GRU networks (“Gate Recurrent Unit”) for example.

[0102] The input data of the neural network are normalized between 0 and 1 by dividing each of the values ​​by a predetermined value (e.g. 10, 100 or 1000) or by the maximum value of the values ​​calculated on all the input data.

[0103] Each phase of the invention is now described in more detail.

[0104] Figure 3 represents, on a flowchart, the steps for implementing a method for learning an artificial intelligence model according to an embodiment of the invention.

[0105] The method begins with a step 301 of initializing the parameters of the model to be trained. More precisely, the parameters are initialized in order to learn a function f that allows calculating the weighting coefficients { associated with measures {p.} from a set of metrics that will be 1 ÏJ(=1JV explained later.

[0106] Step 302 consists of generating a training database to perform the model training. The training data consists of several sets, each set consisting of several radio measurements, of a solution associated reference positioning as well as a 3D map of the environment including the reference position of the node to be located, the positions of the terrestrial base stations (reference nodes) as well as all the obstacles in the environment.

[0107] The database can be built by collecting real {pp measurements by means of a receiver and by measuring the reference position from a high-performance system (x^A z^ ), for example an inertial system or a high-precision GNSS system or a combination of the two systems.

[0108] The measurements {p are either time-of-flight measurements of the transmitted signal between the mobile node to be located and a reference node (a base station), or distance measurements obtained by multiplying the time-of-flight measurements by the wave speed.

[0109] The measurements may also include C / N0 signal-to-noise ratio indicators.

[0110] According to another embodiment, the training database is generated by simulation.

[0111] Two sets of metrics are calculated from the training database.

[0112] A first set of joint metrics is calculated in the form of a set of residuals organized in an MR matrix as shown in [Fig.4].

[0113] Residuals are joint metrics that take into account the simultaneous impact of multiple measurements on several communication links between the mobile node to be located on the one hand and several reference nodes.

[0114] The MR matrix of residuals is constructed from A7 measurements taken for N distinct reference nodes. The measurements are, for example, time-of-flight measurements Tb. For each row of the matrix, a subset Sn of these measurements is selected, comprising A7 - 1 measurements (all except one): Sn = { i n. For For each of the subsets Sn, we calculate a residual ÿT” for each measure, with i varying from 1 to N, i^n and n varying from 1 to N.

[0115] The residual is equal to T” = T; - , with T; the measurement and T'i the theoretical flight time calculated from the distance between the moving node and the reference node of index i.

[0116] The MR matrix of residues is therefore a square matrix with N rows and N columns.

[0117] In detail, each residual is calculated as follows.

[0118] From the N time-of-flight measurements, we define N sets of measurements each comprising N-1 measures Sn — {T}} . i^n.

[0119] For each set S„, a positioning solution f„ is calculated by means of a least squares solution of the following equations.

[0120] Considering a moving node of position r = [ * JZ ] T and a base station (reference node) of position F _ y, Z / ]' the time-of-flight measurement T; between the ' i lAi Two nodes can be modeled using the following relationships: [YES] Tt = l^, +

[0122] With

[0123] r;)=i^(xx^2^ (yy.)2+ (zz,-)2 ( 11 )

[0124] to is a time synchronization error and v is a random variable, following a Gaussian distribution, which represents measurement errors and transmission channel errors, in particular related to multipath and NLOS propagation.

[0125] A positioning solution r„ is calculated by finding the minimum of the function given in equation (12)

[0126] r„ ^argmin / r (12)

[0127] With

[0128] pA^t 1 »\

[0129]

[0130]

[0131] The term r) corresponds to a residual calculated between the measurement T; and the theoretical flight time £ rj The weighting coefficients are taken to be equal to a predefined value, for example they are all taken to be equal to 1. Next, each residual in the MR residual matrix is ​​calculated using the following relationship:

[0132] ^ = ^-^(^),^=1-^(14)

[0133] The second set of ML link metrics is shown in [Fig.5], it has as many lines as there are reference nodes.

[0134] Each line contains several indicators relating to the quality of the link.

[0135] For example, a first possible indicator is a measurement of the signal-to-noise ratio on the RSS link;

[0136] To complement these indicators, it is proposed to rely on a 3D map of the environment in which each node associated with the different measures provided is positioned, as well as the obstacles, particularly urban ones, present in the area.

[0137] Three types of possible indicators are then determined that influence the propagation of the signal: a first set 501 of indicators characterizing the direct paths between the moving node to be located and a reference node, a second set 502 of indicators characterizing the paths diffracted via obstacle angles between the moving node to be located and a reference node, and a third set 503 indicators characterizing the reflected paths via obstacles, between the mobile node to be located and a reference node.

[0138] Figure 6 illustrates, in a simplified diagram, the different possible paths of a signal transmitted between a mobile node M and a reference node R in an environment including two obstacles Bi,B2 which are for example buildings.

[0139] The direct path 601 between the two nodes Nm, Nref crosses the building Bb II is consequently attenuated by the crossing of this building.

[0140] The first set of indicators 501 characterizing direct paths includes at least one indicator from among: the number of obstacles Nobs crossed by the signal and the total distance D1N traveled by the signal through the obstacles (for example buildings) crossed.

[0141] The number of obstacles traversed by the signal is an indicator of the NLOS nature of the path. A number of obstacles equal to 0 indicates LOS propagation.

[0142] The total distance that the signal travels through buildings also gives information on the strength of the received signal and in particular its attenuation.

[0143] Path 602 is a diffracted path on an angle of building Bb

[0144] The second set 502 of indicators characterizing the diffracted paths includes at least one indicator among: the additional distance Dif fEDj traveled by the diffracted path compared to the direct path 601, the diffraction angle Diff . between the two rays of the diffracted path.

[0145] The DiffEDj distance gives an indication of the positioning error made if the diffracted path is detected as a direct path.

[0146] The diffraction angle provides an indication of the probability that the diffracted path will be received. For example, if this angle is close to 180°, this indicates low diffraction and therefore a high probability of receiving the signal. If the diffraction angle is lower, for example on the order of 100°, this indicates high diffraction associated with significant signal attenuation and a low probability of reception.

[0147] Finally, path 603 is a reflected path on one face of building B2.

[0148] The second set 503 of indicators characterizing the reflected paths includes at least one indicator from among: the additional distance ReflEDi traveled by the reflected path compared to the direct path 601, the reflection angle ^€Pai between the two radii of the reflected path

[0149] The ReflEDi distance gives an indication of the positioning error made if the reflected path is detected as a direct path.

[0150] The angle of reflection gives an indication of the probability that the reflected path will be received.

[0151] The second set of ML metrics is determined from a 3D map of the environment in which all the data necessary for the geometric calculations of the metrics are mapped, namely: the position of the moving node and the reference nodes, as well as the 3D coordinates of the various obstacles. The indicators 501, 502, and 503 of the different paths are thus calculated from geometric information that is independent of the measurements provided but reproduces the same geometric configurations as those associated with the measurements.

[0152] The two sets of metrics MR, ML are then provided as input to the model and, in step 303, for each set of measurements in the training database, a model inference phase is executed to calculate the weighting coefficients from the network parameters.

[0153] In step 304, a set of reference weighting coefficients are determined. Advantageously, they are calculated from the reference position and reference measurements using, for example, one of the following formulas, where is calculated for the reference position of the moving node xref

[0154]

[0155]

[0156] In general, the weighting coefficients are taken to be equal to

[0157] wref —_____with k an integer greater than or equal to 1

[0158] At step 305, the model is trained so as to adjust its parameters {seen} in function : calculated weighting coefficients: | reference weighting coefficients: { ^ref lk

[0159] More specifically, if the model is a neural network, the parameter adjustment is carried out, for example, by means of: a propagation phase of training data in the neural network from the network's inputs to its outputs to calculate weighting coefficients! w.} k , an error calculation between the calculated output and the result to be obtained via a cost function, A backpropagation phase of the error aims to update the synaptic parameters or coefficients for each connection between two neurons using a backpropagation algorithm that minimizes the cost function. The backpropagation algorithm is, for example, based on gradient descent or any other equivalent algorithm.

[0160] The cost function is a distance function between the calculated weighting coefficients and the reference weighting coefficients so as to make the calculated weighting coefficients converge towards the reference coefficients.

[0161] For example, the following cost function can be used in place of the one given in equation (9):

[0162] / / ky* y^ hM* U?ll (15)

[0163] Any other cost function that reflects a difference between the coefficients calculated by the algorithm and the reference coefficients may be used.

[0164] Once the model is trained, it can be used to predict positioning information.

[0165] Fig. 7 represents a flowchart of an example of a method for determining a diffracted path 602 associated with a pair (moving node M, reference node R) taking into account the environment mapped in a 3D map.

[0166] At step 700, the variables dmin = -1, i min = -1, pmin=[0,0,0]T corresponding respectively to the distance traveled by the diffracted path, the index i of the building on which the signal is diffracted and the diffraction point on an edge of the building, are initialized.

[0167] In step 701, it is determined, via the 3D map, whether at least one obstacle is present on the direct path between the moving node of position r and the reference node of position rB.

[0168] If there is at least one obstacle, we proceed to step 702 which consists of identifying all the edges e; of all the obstacles present in the scene.

[0169] At step 703, we calculate the point p; on the edge e; which minimizes the distance ^diff = 11 rB ' P-11 + 11r " P-11 °ù HH denotes the norm of a vector.

[0170] At step 704, we detect if one of the two segments [rB, Pi] or [r, p;] crosses an obstacle, if so, this path is not taken into account, otherwise we proceed to the next step.

[0171] At step 705, we check if the calculated distance d^ff is less than the saved distance value; if not, we loop back to step 702 to move to the next edge; if so, we proceed to step 706 to save the current values: d min = d diff A min 1? Pmin- pi

[0172] We loop back to step 702 until all edges of all obstacles have been checked, then we proceed to the final step 707 to provide as output the values ​​d min = d diff rl min 1? Pmin- pi •

[0173] In one embodiment, the iterations of steps 702 to 706 are limited to the edges of obstacles located respectively in a first predefined spatial zone including the emitter and in a second predefined spatial zone including the receiver. This embodiment has the advantage of reducing the computational cost of the algorithm by assuming that reflections / diffractions / obstructions via obstacles far from the emitter or receiver are less probable than those from surrounding obstacles close to the two nodes.

[0174] From these values, one can then calculate the difference between the distance dmin of the selected diffracted path and the distance || rB - r || of the direct path. One can also calculate the diffracted angle from the respective coordinates of the points rB, r and p;.

[0175] Fig. 8 illustrates an example of a method for determining a reflected path 603 associated with a pair (moving node M, reference node R) taking into account the environment mapped in a 3D map.

[0176] At step 800, the variables dmdl(k) = -1, i min(k) = -1, pmin(k) = [0, 0, 0]T, k = l are initialized.

[0177] The index k varies from 1 to Nreflect, which is equal to the number of reflected paths that we wish to model. This number is, for example, equal to 2. It is a parameter of the method.

[0178] At step 801, we iterate Nrefiect times the following steps.

[0179] At step 802, all the faces f; of all the obstacles in the scene are determined.

[0180] In step 803, the point p is determined on the face £ which minimizes the distance drefI = 11 rB - p. 11 + ||r- p. 11 where II denotes the norm of a vector.

[0181] In step 804, we detect if one of the two segments [rB, p;] or [r, p;] crosses an obstacle, if so, this path is not taken into account, otherwise we proceed to the next step.

[0182] At step 805, we check if the calculated distance dreîi is less than the saved distance value; if not, we loop back to step 802 to move to the next face; if so, we proceed to step 806 to save the current values: Jm,„(k) — d diff ii min (k)— i, Pmin(k)- pi

[0183] We loop back to step 802 until all faces of all obstacles have been checked, then we proceed to step 807 where the face fimin is removed from the list of faces, and we loop back to step 801 until k has reached Nrefiecf

[0184] Finally, in the final step 808, the outputs d min, i min: pmin are provided.

[0185] From these values, one can then calculate the difference between the distance dmin of each reflected path and the distance 11 rB - rj| of the direct path. One can also calculate the angle of reflection from the respective coordinates of the points rB, r and p;.

[0186] Figure [Fig.9] represents a flowchart of a method, according to an embodiment of the invention, for locating a mobile node, assisted by the artificial intelligence model trained by means of the method described in Figure [Fig.3].

[0187] The method begins with a step 900 of receiving a set of time-of-flight measurements {TJ (or pseudo-distances) for i ranging from 1 to N where N is the number of reference nodes (base stations in the case of a terrestrial network). Each measurement is characteristic of the time of flight between the same mobile node to be located and one of the reference nodes.

[0188] Optionally, the method also receives as input quality information from each RSS radio link;.

[0189] It also receives a three-dimensional map C of the communications network environment as well as a set of position hypotheses L which are, for example, predefined positions on a given resolution grid of the map C. In other words, the map C is decomposed into cells, each cell being associated with a position hypothesis.

[0190] In step 901, the first set of joint MR metrics is calculated from the time-of-flight measurements in the form of a residual matrix as illustrated in [Fig.4].

[0191] Each residual is calculated in the manner described above to construct the ML matrix.

[0192] In step 902, the second set of ML link metrics is determined from the map C and each position hypothesis *k.

[0193] Metrics 501, 502, and 503 are calculated in the manner described above in support of [Fig. 5] for each positional hypothesis. In other words, several second sets of MLj link metrics are obtained.

[0194] In one embodiment, the second set of metrics further includes other metrics such as metrics characterizing the signal-to-noise ratio of the link or the elevation angle.

[0195] For each position hypothesis I'a, the first set of MR metrics and the second set of MLj metrics are provided as input to the artificial intelligence model trained during the training phase.

[0196] Step 903 consists of performing a model inference phase from said metrics. The model provides as output a set of optimal weighting coefficients "j, ...

[0197] In step 904, a new positioning solution is calculated by finding the position coordinates that minimize the following function:

[0198] a , \ _ oroTninYV (1$)

[0199] ÔT-(x) is the residual calculated by taking the difference between the time-of-flight measurement T; and the theoretical time-of-flight traveled by the signal between the node of position rk and the reference node of index i of known position.

[0200] Steps 903 and 904 are executed for each position hypothesis associated with map C.

[0201] Finally, in step 905, a consistency score is calculated for each pair (position hypothesis *k, calculated position r^.).

[0202] The score is, for example, equal to:

[0203] Sa>ret = ph(tk, t <a,tt ...,rv)+                   <17) 102041 with p.Jfy = Ud)2

[0205] £Ojt is the synchronization error which is also obtained by solving equation (16).

[0206] Po is a 3x3 matrix which is used to take into account the error tolerances between the tested positions fk and the calculated positions rk.

[0207] For example,

[0208] c?2 0 0 ' pQ = 0 0 0 0 ¢72.

[0209] The coefficients °x, and are fixed at values ​​which reflect the margins of error on the three coordinates.

[0210] For example, <7X — 1m, ay = 2m and (ïz = 3m.

[0211] If the difference between the two positions (tested and calculated) is less than these tolerances, this results in a high score, otherwise the score decreases according to the error.

[0212] Once all Scorek scores have been calculated, the final position of the mobile node is determined from the calculated scores, for example by selecting the tested position that provides the highest score or by averaging the score-weighted positions using the following formula:

[0213] has €1. , with K the number of positions tested on the map. - EU™".

[0214] The localization method according to the invention can be executed in a computing processor of a radio receiver or a satellite radionavigation receiver embedded in the mobile node to be located.

[0215] Alternatively, the localization method can be executed in a remote server from measurements made by said receiver and then transmitted to that server.

[0216] In general, the learning and localization methods according to the invention can be implemented using hardware and / or software components. The software components may be available as a computer program product on a computer-readable medium, which may be electronic, magnetic, optical, or electromagnetic. The hardware components may be available in whole or in part, including, for example, as dedicated integrated circuits (ASICs) and / or field-configurable integrated circuits (FPGAs) and / or as neural circuits or as a digital signal processor (DSP) and / or as a graphics processing unit (GPU), and / or as a microcontroller and / or as a general-purpose processor.

[0217] The invention is particularly suited to time-of-flight measurements over relatively large distances (typically greater than 100 meters), which relates to LPWAN (Low Power Wide Area Network) type networks, such as the LoRaWAN network, or satellite radionavigation systems (GPS, Galileo) or 4G / 5G type cellular networks.

Claims

1. Demands A computer-implemented method for learning an artificial intelligence model intended to be used to determine the position of a moving node in a given area, the method comprising the steps of: - Receive (302) a training dataset comprising several sets of radio frequency measurements corresponding to signals exchanged respectively between a mobile node of known position and several reference nodes of known positions and a 3D map representing the environment of the mobile node and the reference nodes, - Determine, for each set of measurements, a set of metrics comprising at least a first set (MR) of residuals calculated for several subsets of measurements, each excluding at least one measurement from the set, and a second set (ML) of indicators of radio propagation paths between the mobile node and each reference node, taking into account environmental obstacles represented in the 3D map, - For each set of measurements, i. Determine (304) a set of reference weighting coefficients, ii. Train (303,305) the artificial intelligence model to produce a set of weighting coefficients from the metric sets of the training data, so as to minimize a distance between said weighting coefficients and the reference weighting coefficients, - the weighting coefficients being intended to weight a set of residuals, equal to a difference between a measurement and a predicted positioning information, during a calculation of prediction of a positioning information by minimizing the sum of the residuals squared and weighted by the weighting coefficients.

2. Method of learning an artificial intelligence model according to claim 1 wherein each reference weighting coefficient is a function of a residual calculated from each radio frequency measurement and the distances or time of flight between the moving node and each reference node.

3. A method for learning an artificial intelligence model according to any one of the preceding claims, wherein the artificial intelligence model is an artificial neural network, for example a recurrent neural network.

4. Method for locating a mobile node in a given area comprising the steps of: - Receiving a set of N radio frequency measurements corresponding to signals exchanged respectively between a mobile node to be located and N reference nodes of known positions in the area, N being an integer at least equal to 4, - Receiving a 3D map representing said area, - Determining (901), from the received measurements, a first set of residuals (MR) calculated for several subsets of measurements each comprising at most Nl measurements, - For a plurality of given positions of the map, i. Determining (902), from the 3D map, a second set (ML) of indicators of the radio propagation paths between said position of the map and the position of each reference node, taking into account the obstacles of the environment represented in the 3D map, ii.Execute (903) an inference phase of the artificial intelligence model trained by means of the learning method according to any one of the preceding claims, from the first and second set to determine a set of weighting coefficients, iii. Determine (904) a provisional position of the moving node from said measurements and the weighting coefficients by finding the value of the. position that minimizes the sum of the squared residuals associated with the N measures, and weighted by the weighting coefficients, iv. Calculate (905) a consistency score between the provisional position and the given position of the map, and - Determine the final position of the moving node as a function of the consistency score.

5. Method of locating a moving node according to claim 4 wherein the consistency score is calculated from the sum of the squared residuals associated with the N measurements, and weighted by the weighting coefficients, the residuals being calculated for the provisional position.

6. A method for locating a moving node according to any one of claims 4 or 5 wherein the final position of the moving node is the position on the map that gives the highest consistency score or is equal to an average of the positions on the map weighted by their respective consistency scores.

7. Method according to any one of the preceding claims wherein the second set (ML) of radio propagation path indicators comprises at least one indicator from among a direct path indicator, a diffracted path indicator and a reflected path indicator.

8. Method according to claim 7 wherein at least one direct path indicator is taken from an indicator of the number of buildings intercepted by the path or an indicator of the total distance traveled by the signal inside the buildings traversed.

9. Method according to claim 7 wherein at least one diffracted path indicator is taken from an indicator of the additional distance traveled for the diffracted path compared to the direct path or an indicator of the diffraction angle.

10. Method according to claim 9 comprising determining the diffracted path as the minimum distance path intercepting an edge of an obstacle among all the obstacles present on the 3D map.

11. Method according to claim 7 wherein at least one reflected path indicator is taken from a distance indicator additional distance traveled for the reflected path compared to the direct path or an indicator of the angle of reflection.

12. Method according to claim 11 comprising determining the reflected path as being the path of minimum distance reflected on a face of an obstacle among all the obstacles present on the 3D map.

13. Method according to any one of the preceding claims wherein the radio frequency measurement is a time-of-flight measurement and a residual is calculated as the difference between the measurement and the theoretical time-of-flight between the position of the moving node and the position of a reference node or the radio frequency measurement is a distance measurement and a residual is calculated as the difference between the measurement and the distance between the moving node and the reference node.

14. Method according to any one of the preceding claims wherein the radio frequency measurements are satellite radionavigation measurements and the reference nodes are satellites or the reference nodes are ground stations.

15. A mobile node localization system comprising a processing unit configured to execute the steps of the localization method according to any one of claims 5 to 14.

16. Computer program comprising instructions for carrying out the method according to any one of claims 1 to 14, when the program is executed by a processor.

17. Processor-readable recording medium on which is recorded a program containing instructions for executing the method according to any one of claims 1 to 14, when the program is executed by a processor.