Method for recognizing or identifying a radar transmitter, device and associated computer program product

Abstract

Method for recognizing or identifying a radar transmitter, associated device and computer program product. The present invention relates to a method for recognizing or identifying a radar transmitter, comprising the following steps: - receiving a radar signal by a receiver (12), - segmenting the received signal in time and frequency, - demodulating the segmented signal, - extracting a rise edge from the demodulated signal, - calculating the real and imaginary parts of the rise edge as a function of time, - defining a path associated with the signal, - scaling the path so that the scaled path is invariant to an initial phase of the received signal, and - comparing the path associated with the scaled signal with at least one reference path, independently of the travel speeds of said paths. The invention further relates to an associated device and computer program product. Figure for the abstract: Fig 1

Description

Title of the invention: Method for recognizing or identifying a radar transmitter, device and associated computer program product

[0001] The present invention relates to a method for recognizing or identifying a radar transmitter, and an associated device.

[0002] Waveforms are increasingly defined by software.

[0003] Such a feature offers the possibility, for the same radar, of easily changing waveforms, or even of parameterizing them in real time. This makes it possible, in particular, to adapt to environmental conditions and / or to make characterization and tracking more difficult for an unwanted listening system.

[0004] This also renders obsolete identification methods based on comparing the waveform, for example frequency channels or time periodicities of radar pulses, with a library obtained previously from previous observations.

[0005] Furthermore, radar transmitters are likely to have several operating modes. Restricting the use of certain operating modes makes it possible to ensure that this operating mode is little or not referenced in libraries, so as to avoid any subsequent recognition.

[0006] The aim of the invention is then to propose a method for recognizing or identifying a radar transmitter that is not sensitive to changes in waveforms.

[0007] To this end, the invention relates to a method for recognizing or identifying a radar transmitter, comprising the following steps:

[0008] - reception of a radar signal by a receiver,

[0009] - segmentation of the received signal in time and frequency,

[0010] - demodulation of the segmented signal,

[0011] - extraction of a rising edge of the demodulated signal,

[0012] - calculation of the real and imaginary parts of the rise front as a function of time,

[0013] - definition of a path associated with the signal, the path being a curve in a plane Cartesian, in which the real part of the rising front is represented on the x-axis and the imaginary part of the rising front is represented on the y-axis,

[0014] - scaling the path, so that the scaled path is invariant to an initial phase of the received signal, and

[0015] - comparison of the path associated with the scaled signal with at least one path reference regardless of the travel speeds of said paths.

[0016] The signal is associated here with the propagation of a path in the complex plane, this path being characterized so as to be invariant under reparameterization, that is to say, the characterization of the curve is identical regardless of the speed at which the path is traversed. Now, the change in the waveform corresponds precisely to a reparameterization, and not to a change in the path. The method therefore makes it possible to recognize or identify a radar transmitter, even if the waveform has been reparameterized.

[0017] According to other advantageous aspects of the invention, the method comprises one or more of the following features, taken individually or in all technically possible combinations:

[0018] - at least one reference path includes paths associated with signals previously received, and wherein, if during the comparison step, the path associated with the signal is similar to at least one of the paths associated with previously received signals, then the method includes a step of classifying the received signal with the at least one previously received signal whose associated path is similar;

[0019] - at least one reference path is associated with a known radar transmitter, and, in which, if during the comparison step, the path associated with the signal is similar to at least one reference path, then the method includes a step of identifying the transmitter of the received signal, the transmitter being the known radar transmitter of the similar reference path;

[0020] - the method includes a step of filtering and subsampling the signal segmented at a filter frequency upstream of the demodulation stage;

[0021] - the path scaling step includes scaling the path of so that the path ends at a given point on the Cartesian plane;

[0022] - the demodulated signal exhibits a power plateau, the rise edge comprising the segments of the received demodulated signal having a power between a minimum value and a maximum power value, the minimum power value being between 10% and 30% of the power plateau, the maximum power value being between 80% and 95% of the power plateau;

[0023] - the method includes a step of defining a signature of the path put to the scale, the curve corresponding to a continuous function in the Cartesian plane, the signature comprising the sequence of coefficients obtained by iterated Riemann-Stieltjes integrations over the coordinates of the points of the function, the step of comparing the path associated with the signal comprising a step of comparing the signature of said path or of a parameter linked to the signature with the signature or the linked parameter of each reference path; and / or

[0024] - the step of comparing the path associated with the signal includes a step of dynamic time warping to measure the similarity between paths.

[0025] The invention also relates to a device for recognizing or identifying a radar transmitter, comprising a radar receiver, a segmentation module, a demodulation module, an extraction module, a module for calculating real and imaginary parts, a path definition module, a scaling module, and a comparison module, the device being capable of implementing the method as defined above.

[0026] According to other advantageous aspects of the invention, the device includes the following feature: the radar receiver includes a superheterodyne receiver.

[0027] The invention also relates to a computer program product comprising software instructions which, when executed by a computer, implement the steps of segmentation, demodulation, rise edge extraction, calculation of the real and imaginary parts, definition of an associated path, scaling of the path, and comparison of the process as defined above.

[0028] The invention will become clearer upon reading the following description, given solely by way of non-limiting example, and made with reference to the drawings in which:

[0029] [Fig-1] [Fig.1] is a schematic representation of a device recognition or identification of a radar transmitter according to an embodiment of the invention,

[0030] [Fig.2] [Fig.2] is a schematic representation of a method for recognizing or identifying a radar transmitter according to an embodiment of the invention,

[0031] [Fig.3] [Fig.3] is an example of power and phase curves associated with an emitter having an amplitude control x, and

[0032] [Fig.4] [Fig.4] is an example of paths scaled according to the method of the invention.

[0033] The invention relates to a method 110 and a device 10 for recognizing or identifying a radar transmitter, an embodiment of which will be described with reference to figures 1 and 2.

[0034] The method 110 for recognizing or identifying a radar transmitter comprises the following steps:

[0035] - reception 112 of a radar signal r by a receiver,

[0036] - segmentation 114 of the received signal in time and frequency,

[0037] - optionally, filtering and subsampling 116 of the segmented signal,

[0038] - demodulation 118 of the segmented signal,

[0039] - extraction 120 of a rising edge of the demodulated signal,

[0040] - calculation 122 of the real and imaginary parts of the rising front as a function of time,

[0041] - definition 124 of a path associated with the signal, the path being a curve in a Cartesian plane, in which the real part of the rising front is represented on the abscissa and the imaginary part of the rising front is represented on the ordinate,

[0042] - scaling 126 of the path allowing one-phase path invariance initial of the received signal, and

[0043] - comparison 128 of the path associated with the scaled signal with at least one reference path regardless of the travel speeds of said paths.

[0044] A radar transmitter emits a signal.

[0045] The emitted signal has the following complex representation:

[0046] [Math.l] s(t) = ]jy[x(t)) xe^CO)

[0047] with x the transmitter amplification control, t the time variable, y(x) the power, and <p(x) le déphasage du signal produit en sortie de l’amplification de l’émetteur radar.

[0048] The time variable t varies here between 0 and a front duration, for example equal to 100 ns.

[0049] An example of power and phase curves associated with an emitter as a function of the amplitude control x is given as an example in [Fig.3].

[0050] By observing the output power of a transmitter with respect to the amplification control, we observe unintentional amplitude modulations, which are specific to the component used for the amplification of the signal to be emitted, and therefore to the transmitter, and in phase.

[0051] Each radar transmitter has a compression point, corresponding to the point where the amplification control stops at a peak input power such that the output power is reduced by a given factor compared to the linear regime.

[0052] The receiver then receives a signal r corresponding to the emitted signal attenuated by a propagation factor a and to an initial phase q>0 up to and to which is added a measurement noise w(t), the measurement noise being for example an additive Gaussian noise.

[0053] The received signal r can then be written as follows:

[0054] [Math.2] r(t) = ae^stt) + w(t).

[0055] The receiver here includes an antenna, the antenna being capable of receiving a signal in a reception band.

[0056] The received signal is then, for example, filtered, amplified and frequency transposed with a mixer, in a known manner.

[0057] Filtering, amplification, and transposition are, for example, implemented by a superheterodyne receiving unit included in the receiver.

[0058] Filtering makes it possible in particular to reduce measurement noise w(t).

[0059] The received signal is here digitized in a sub-band of the frequency spectrum, for example with a width between 500 MHz and 1 GHz, for example by a digitizer.

[0060] The sub-band is included in the receive band.

[0061] The digitizer here samples the channels in phase and in quadrature with a sampling frequency higher than the receive band.

[0062] Then, the signal is segmented in time and frequency.

[0063] The signal is, for example, isolated in time and frequency.

[0064] This makes it possible to isolate the signal associated with a single pulse in the whole of the received signal, which includes, for example, generally several pulses.

[0065] Segmentation is, for example, carried out in a conventional manner.

[0066] The signal is, for example, sent to a data processing unit.

[0067] When a radar pulse is detected, i.e. when a radar signal is detected, here by the data processing unit, then its arrival time and carrier frequency are measured.

[0068] The carrier frequency is, for example, estimated. This estimate is obtained, for example, by searching for the maximum power in a fast Fourier transform.

[0069] From the arrival time and the carrier frequency, a part of the signal, hereafter called the signal, is extracted around its arrival time: the signal is then isolated in time.

[0070] In addition, the signal is, for example, isolated in frequency, for example by means of a filtering step, by means of the carrier frequency information.

[0071] In a particular embodiment, the process then includes a filtering and subsampling step 116 of the segmented signal at a filter frequency upstream of the demodulation step.

[0072] Said step is, for example, optional depending on the scanning rate of the receiver, in particular, for example, when the scanning rate already has a filtering effect.

[0073] This step makes it possible in particular to improve the signal-to-noise ratio by reducing the sampling rate, for example to a value on the order of a few tens of MHz or a hundred MHz.

[0074] Then, the process includes the step of demodulating the signal 118.

[0075] The demodulation step 118 includes, for example, a substep of transposing the signal into a baseband for sampling by double quadrature demodulation or via the analytical signal.

[0076] A carrier residual is defined as equal to the difference between the carrier frequency and the demodulation frequency.

[0077] The demodulation frequency is, for example, here the center frequency of the sub-band in which the received signal is digitized, as described above.

[0078] In a particular embodiment, the method further includes a step for refining the carrier residual estimation. This refinement step includes, for example, a substep for refining the carrier frequency estimation, in addition to the measured value described above, for example using the Fourier transform of the extracted signal samples.

[0079] Then, the rising edge of the demodulated signal is extracted.

[0080] Here, the demodulated signal has a power plateau, the rise edge comprising the segments of the received demodulated signal having a power between a minimum value and a maximum power value, the minimum power value being between 10% and 30% of the power plateau, the maximum power value being between 80% and 95% of the power plateau.

[0081] More specifically, said segments are here extracted from the demodulated signal to form the rise front.

[0082] Here, the carrier residue is, for example, digitally demodulated, so as to extract the rise edge in the baseband.

[0083] The real and imaginary parts of the rising front as a function of time are calculated, that is to say the Cartesian coordinates of the signal corresponding to the rising front.

[0084] Then, the path associated with the signal is defined.

[0085] The path corresponds to the curve in a Cartesian plane, in which the real part of the rising front is represented in abscissas and the imaginary part of the rising front is represented in ordinates.

[0086] The curve here corresponds to a function X(t) continuous in the Cartesian plane, parameterized by the time variable t, for example with t G [0 ;1].

[0087] The method includes scaling the path allowing invariance of the path to an initial phase of the received signal, and furthermore invariance of the path to the propagation factor a.

[0088] The path scaling step 126 includes, for example, scaling the path so that the path terminates at a given point on the Cartesian plane.

[0089] The path starts at point (0; 0).

[0090] Examples of different scaled paths are, for example, given by way of example in [Fig.4].

[0091] In this example, the set of paths has been scaled so as to end at the point (1;0).

[0092] The scaling step 126 is likely to be carried out directly on the path, i.e. on the curve.

[0093] Additionally or alternatively, scaling is performed directly on the rise front or on the real and imaginary parts.

[0094] For example, all signal samples are divided by the last value of the rising edge of said signal.

[0095] Alternatively, the path is characterized by rotationally invariant signatures, so that its characterization is invariant with respect to the initial phase.

[0096] Said characterization includes, for example, a complex division or a rotationally invariant F treatment, such as

[0097] [Math.3] F(z)=F(ze ie ) for any angle 0,

[0098] of all paths.

[0099] The scaling step 126 is, for example, carried out upstream or downstream of the definition step 124.

[0100] The path associated with the scaled signal is then compared with at least one reference path independently of the travel speeds of said paths.

[0101] In a particular embodiment, the method includes a step of defining a signature of the scaled path.

[0102] The signature includes, for example, the sequence of coefficients obtained by iterated Riemann-Stieltjes integrations over the coordinates of the points of the function X.

[0103] With the two coordinates of the points on the curve denoted Xæ(t) ; [0,1] Ket X(2)(t) : [0,1] -» ®., the coefficients of order k > 1 of the signature are given by:

[0104] [Math.4] 5^ = f ... ,

[0105] The coefficients of order k are thus 2^-

[0106] The order k is, for example, chosen according to the desired precision and / or the available computing power.

[0107] Here, the order k is, for example, between 2 and 8.

[0108] The signature of a path has the property of being invariant under re-parameterization, that is to say that the coefficients obtained are identical regardless of the speed of traversing the path - in other words, the curves and X ( i / r ( t ) ) generate the same coefficients St if: [ 0,1] —> [ 0,1] is continuous, and that ( / / (1)-1=1 / / (0)=0.

[0109] In a particular embodiment, a parameter related to the signature is, for example, considered.

[0110] The parameter includes, for example, the log-signature, corresponding to the formal logarithm of the signature in the power series algebra in a given number, here two, indeterminate.

[0111] The signature and log-signature of a path are, for example, partly documented in the article A primer on the signature method in machine learning, Chevyrev, A. Kormilitzin, online: DOI:10.48550 / arXiv.1603.03788.

[0112] The comparison step 128 of the path associated with the signal then includes a step of comparing the signature of said path or of a parameter linked to the signature, here for example the log-signature, with the signature or the linked parameter of each reference path.

[0113] Alternatively, the path-associated signal comparison step 128 includes a dynamic time warping step to measure the similarity between paths, for example as described in the article Alignment of curves by dynamic time warping, K. Wang, T. Gasser, The annals of statistics 25(3) pp. 1251-1276. 1997

[0114] The dynamic time deformation step includes a re-parameterization of the paths in a common time reference frame.

[0115] The scaled path is then compared with at least one reference path independently of the travel speeds of said paths.

[0116] The comparison is, for example, carried out using artificial intelligence, for example by an artificial neural network, in particular trained on reference paths.

[0117] The neural network comprises an ordered succession of layers of neurons, each of which takes its inputs from the outputs of the previous layer.

[0118] More precisely, each layer comprises neurons taking their inputs from the outputs of the neurons of the previous layer, or from the input variables for the first layer.

[0119] Alternatively, more complex neural network structures can be envisaged with a layer that can be linked to a layer further away than the immediately preceding layer.

[0120] Each neuron is also associated with an operation, that is to say a type of processing, to be carried out by said neuron within the corresponding processing layer.

[0121] Each layer is connected to the other layers by a plurality of synapses. A synaptic weight is associated with each synapse, and each synapse forms a link between two neurons. It is often a real number, which takes both positive and negative values. In some cases, the synaptic weight is a complex number.

[0122] Each neuron is designed to perform a weighted sum of the value(s) received from the neurons of the preceding layer, each value being multiplied by the respective synaptic weight of each synapse, or connection, between said neuron and the neurons of the preceding layer, and then to apply an activation function, typically a non-linear function, to said weighted sum, and to deliver at the output of said neuron, in particular to the neurons of the next layer connected to it, the value resulting from the application of the activation function. The activation function introduces non-linearity into the processing performed by each neuron. The sigmoid function, the hyperbolic tangent function, and the Heaviside function are examples of activation functions.

[0123] As an optional complement, each neuron is also capable of applying, in addition, a multiplicative factor, also called bias, to the output of the activation function, and the value delivered at the output of said neuron is then the product of the bias value and the value from the activation function.

[0124] A convolutional neural network is also sometimes called a convolutional neural network or by the acronym CNN, which refers to the English term "Convolutional Neural Networks".

[0125] In a convolutional neural network, each neuron in the same layer has exactly the same connection pattern as its neighboring neurons, but at different input positions. The connection pattern is called the convolutional kernel or, more often, the "kernel" in reference to the corresponding English term.

[0126] A fully connected layer of neurons is a layer in which the neurons of said layer are each connected to all the neurons of the preceding layer.

[0127] Such a layer is more often referred to by the English term "fully connected", and sometimes designated as a "dense layer".

[0128] In one embodiment, at least one reference path includes paths associated with previously received signals.

[0129] If during the comparison step 128, the path associated with the signal is similar to at least one of the paths associated with previously received signals, then the method includes a step of classifying the received signal with the at least one previously received signal whose associated path is similar.

[0130] The process then includes, for example, a step of recognizing the transmitter that emitted the received signal, that is to say that the pulses from said transmitter are classified together.

[0131] The method according to the invention makes it possible to carry out such recognition independently of the waveform used by the transmitter.

[0132] Additionally or alternatively, at least one reference path is associated with a known radar transmitter.

[0133] If during the comparison step 128, the path associated with the signal is similar to at least one reference path, then the method includes a step of identifying the transmitter of the received signal, the transmitter being the known radar transmitter of the similar reference path.

[0134] In a particular embodiment, the at least one reference path comprises a plurality of reference paths, each reference path being associated with a known radar transmitter from a list of known transmitters.

[0135] The at least one reference path includes at least one associated path for each known emitter in the list.

[0136] The comparison step 128 then includes comparing the path associated with the scaled signal with at least one reference path, so as to determine the known transmitter from the list that most likely emitted the received signal.

[0137] The process then includes a step of determining the transmitter that most likely emitted the received signal, corresponding to the known transmitter in the list that most likely emitted the received signal.

[0138] The comparison thus makes it possible to see the similarities between the path of the received signal and reference paths, to group similar paths and / or to recognize a probable transmitter of the received signal, without this comparison being sensitive to changes in waveforms.

[0139] The invention further relates to an electronic device for recognizing or identifying a radar transmitter, capable of implementing the recognition or identification process as described above.

[0140] The electronic device 10 includes a receiving module or receiver 12, a segmentation module 14, a demodulation module 18, an extraction module 20, a calculation module 22 of the real and imaginary parts, a path definition module 24, a scaling module 26, and a comparison module 28, as shown in [Fig. 1].

[0141] As an optional complement, the electronic device 10 includes a filtering and subsampling module 16.

[0142] The electronic device 10 includes an information processing unit formed for example of a memory and a processor associated with the memory.

[0143] In the example of [Fig. 1], the receiver module 12, the segmentation module 14, the demodulation module 18, the extraction module 20, the real and imaginary part calculation module 22, the path definition module 24, the scaling module 26, and the comparison module 28, as well as, optionally, the filtering and subsampling module 16, are each implemented as software, or a software component, executable by the processor. The memory of the electronic device 10 is thus capable of storing receiver software, segmentation software, demodulation software, extraction software, real and imaginary part calculation software, path definition software, scaling software, and comparison software, as well as, optionally, filtering and subsampling software. The processor is then capable of running each of the above-mentioned software programs.

[0144] In an alternative not shown, the receiver module 12, the segmentation module 14, the demodulation module 18, the extraction module 20, the calculation module 22 for real and imaginary parts, the path definition module 24, the scaling module 26, and the comparison module 28, as well as, optionally, the filtering and subsampling module 16, are each implemented as a programmable logic component, such as an FPGA (Field Programmable Gate Array), or as an integrated circuit, such as an ASIC (Application-Specific Integrated Circuit).

[0145] When the electronic device 10 is implemented in the form of one or more software programs, that is, in the form of a computer program, also called a computer program product, it is further capable of being stored on a computer-readable medium, not shown. The computer-readable medium is, for example, a medium capable of storing electronic instructions and being connected to a bus of a computer system. By way of example, the readable medium is an optical disc, a magneto-optical disc, ROM, RAM, any type of non-volatile memory (for example, FLASH or NVRAM), or a magnetic card. A computer program comprising software instructions is then stored on the readable medium.

[0146] The invention further relates to a computer program product comprising software instructions which, when executed by a computer, implement the steps of the process as described above, apart from the receiving step.

Claims

Demands

1. A method for recognizing or identifying a radar transmitter, comprising the following steps: - receiving (112) a radar signal by a receiver (12), - segmenting (114) the received signal in time and frequency, - demodulating (118) the segmented signal, - extracting (120) a rise edge from the demodulated signal, - calculating (122) the real and imaginary parts of the rise edge as a function of time, - defining (124) a path associated with the signal, the path being a curve in a Cartesian plane, in which the real part of the rise edge is represented on the abscissa and the imaginary part of the rise edge is represented on the ordinate, - scaling (126) the path, so that the scaled path is invariant to an initial phase of the received signal, and - comparing (128) the path associated with the scaled signal with at least one reference path independently of the travel speeds of said paths.

2. A method according to claim 1, wherein the at least one reference path comprises paths associated with previously received signals, and wherein, if during the comparison step (128), the path associated with the signal is similar to at least one of the paths associated with previously received signals, then the method comprises a step of classifying the received signal with the at least one previously received signal whose associated path is similar.

3. A method according to claim 1 or 2, wherein at least one reference path is associated with a known radar transmitter, and wherein, if during the comparison step (128), the path associated with the signal is similar to at least one reference path, then the method includes a step of identifying the transmitter of the received signal, the transmitter being the known radar transmitter of the similar reference path.

4. A method according to any one of claims 1 to 3, comprising a filtering and subsampling step (116) of the segmented signal at a filter frequency upstream of the demodulation step.

5. A method according to any one of claims 1 to 4, wherein the path scaling step (126) comprises scaling the path so that the path terminates at a given point in the Cartesian plane.

6. A method according to any one of claims 1 to 5, wherein the demodulated signal has a power plateau, the rise edge comprising the segments of the received demodulated signal having a power between a minimum value and a maximum power value, the minimum power value being between 10% and 30% of the power plateau, the maximum power value being between 80% and 95% of the power plateau.

7. A method according to any one of claims 1 to 6, comprising a step of defining a signature of the scaled path, the curve corresponding to a continuous function in the Cartesian plane, the signature comprising the sequence of coefficients obtained by iterated Riemann-Stieltjes integrations over the coordinates of the points of the function, the step of comparing the path associated with the signal comprising a step of comparing the signature of said path or of a parameter linked to the signature with the signature or the parameter linked to each reference path.

8. A method according to any one of claims 1 to 7, wherein the path-associated signal comparison step (128) includes a dynamic time-warping step to measure the similarity between the paths.

9. Device (10) for recognizing or identifying a radar transmitter, comprising a radar receiver (12), a segmentation module (14), a demodulation module (18), an extraction module (20), a calculation module (22) for real and imaginary parts, a path definition module (24), a scaling module (26), and a comparison module (28), the device (10) being capable of implementing the method according to any one of claims 1 to 8.

10. Device according to claim 9, wherein the radar receiver (12) comprises a superheterodyne receiver.

11. Computer program product comprising software instructions which, when executed by a computer, implement the steps of segmentation (114), demodulation (118), extraction (120) of a rise front, calculation (122) of the real and imaginary parts, definition (124) of an associated path, scaling (126) of the path, and comparison (128) of the method according to any one of claims 1 to 8.

Images