Combination path echo based non-line-of-sight target positioning and velocity estimation method

Abstract

The application discloses a non-line-of-sight target positioning and speed estimation method based on a combined path echo, first processes radar signals, obtains a distance-Doppler spectrum of a target from a radar original echo, then obtains distance, angle and radial speed information of the target through the distance-Doppler spectrum, then obtains a virtual target in a non-line-of-sight scene in combination with known building scene layout information, classifies virtual targets of different paths through a clustering algorithm, completes multi-path echo classification, obtains a direct positioning result, finally fuses multi-path position information to obtain a real position of the target, simultaneously combines multi-path speed information to obtain a real speed of the target, realizes positioning and speed estimation of the non-line-of-sight target, and the method can realize high-precision positioning of a non-line-of-sight area target, estimates a target speed parameter, guarantees a non-line-of-sight scene target positioning and speed estimation effect, and provides a powerful guarantee for target monitoring technology in a non-line-of-sight area.

Description

Technical Field

[0001] This invention belongs to the field of non-line-of-sight target detection technology, specifically relating to a non-line-of-sight target localization and velocity estimation method based on combined path echoes. Background Technology

[0002] Non-line-of-sight target detection technology is mainly based on the characteristics of electromagnetic waves propagating through multiple paths, such as reflection and diffraction on the surface of media like walls. This enables the effective detection, location, and identification of obscured targets in blind spots such as urban street corners and areas blocked by vehicles, and has significant application value in fields such as intelligent driving. Due to the diversity of the built environment and the complexity of electromagnetic propagation phenomena, many different propagation paths exist when electromagnetic waves propagate between radar and a target. The propagation path from radar to the target directly is called the direct path. The propagation path from radar to the target after one or more reflections on the wall surface is called the multipath path, or simply multi-reflection path. Based on the number of reflections on the wall surface, it can be divided into single-reflection or multi-reflection paths. The propagation path from radar to the target after diffraction is called the diffraction path. When a target is outside the line-of-sight range, since the target is not within the radar's direct line of sight, the electromagnetic waves can only reach the target via one or more reflection paths and diffraction paths. In this case, the detection and location of the target requires the use of electromagnetic waves that have propagated through reflection and diffraction paths. In reality, the complexity of electromagnetic propagation and the diversity of architectural scenes often lead to complex multipath signals. Therefore, methods for locating non-line-of-sight targets based on multipath signals are of great research value.

[0003] Many research institutions both domestically and internationally have conducted research on non-line-of-sight target localization methods based on multipath signals. The literature "JOHANSSON T, ANDERSSON..." GUSTAFSSON M, et al. Positioning of moving non-line-of-sight targets behind a corner. European Radar Conference (EuRAD), London, UK, 2016: 181-184. proposed a non-line-of-sight positioning method based on the first multi-path signal, using the distance value corresponding to the non-line-of-sight target first multi-path reflection obtained by radar, and then realizing the positioning of the non-line-of-sight moving target in the L-shaped building corner scene by means of the geometric symmetry relationship of the first mirror reflection. Literature "GUO Shisheng, ZHAO Qingsong, CUI Guolong, et al. Behind corner targets location using small aperture millimeter wave radar in NLOS urban environment. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020" proposed a method of using the distance and angle information of the first multi-path signal for non-line-of-sight positioning, based on the characteristics of high angle resolution of millimeter wave radar, after obtaining the distance corresponding to the first reflection multi-path, the angle information obtained by the phase comparison method is used to obtain the real position of the target by combining the geometric symmetry relationship of the first mirror reflection. At present, the existing multi-path positioning method represented by the above method mainly uses the first reflection path of the target, and the positioning accuracy is not accurate enough, and in the estimation of the target speed parameter, the above method can only get the radial velocity of the moving target on the first reflection path, and cannot get the overall speed of the target. Therefore, it is of great value to study a non-line-of-sight positioning and speed estimation method based on combined path echoes. SUMMARY

[0004] To solve the above technical problems, the present application provides a non-line-of-sight target positioning and speed estimation method based on combined path echoes, which can realize high-precision positioning of non-line-of-sight area targets and estimate the target speed parameter.

[0005] The technical scheme adopted by the present application is as follows: a non-line-of-sight target positioning and speed estimation method based on combined path echoes, the specific steps are as follows:

[0006] S1, process the radar signal to obtain the distance-Doppler spectrum of the target from the radar original echo, and then obtain the distance, angle and radial velocity information of the target through the distance-Doppler spectrum;

[0007] S2, in a non-line-of-sight scene, combining known building scene layout information, based on step S1, a virtual target is obtained, and the virtual targets of different paths are classified through a clustering algorithm to complete multi-path echo classification and obtain a direct positioning result;

[0008] S3, based on step S2, the target real position is obtained by fusing the multi-path position information, and the target real speed is obtained by combining the multi-path speed information, realizing the positioning and speed estimation of the non-line-of-sight target.

[0009] Further, the step S1 is specifically as follows:

[0010] It is assumed that both the simulation and the measured scene use a millimeter wave radar, the signal transmitted by the radar is a linear frequency modulation signal, the carrier frequency of the radar is f c , the signal bandwidth is B, the signal period is T c , the signal amplitude is A T , the signal phase is , and the signal transmission time is t. The expression of the radar transmission signal x T (t) is as follows:

[0011]

[0012] The transmitted signal encounters a target and reflects a return signal, which is received by the receiving antenna. The expression of the received signal x R (t) is as follows:

[0013]

[0014] Wherein, A R represents the amplitude of the received signal, and τ represents the time delay of the signal, i.e. the time from the radar to the target and back to the radar. In a multi-path environment, the expression is as follows:

[0015]

[0016] Wherein, R represents the distance from the radar to the target, and c represents the speed of light.

[0017] After the receiving antenna receives the return signal reflected by the target, the transmitted signal is mixed to obtain a difference frequency signal y(t), also known as an intermediate frequency signal. The phase of the intermediate frequency signal is the phase difference between the return signal and the transmitted signal, and the expression is as follows:

[0018]

[0019] Wherein, α represents the signal amplitude coefficient, represents the intermediate frequency signal phase, represents the high-order term about t generated in the mixed signal, which can be ignored, and f bThe frequency of the intermediate frequency signal is expressed as follows:

[0020]

[0021] After obtaining the intermediate frequency signal, the intermediate frequency signal is subjected to fast Fourier transform (FFT) processing to obtain the frequency f of the intermediate frequency signal b , and the distance R of the target to the radar is further obtained. The original echo signal collected by the radar is the intermediate frequency signal after mixing, and then a distance FFT operation is performed, that is, the original echo signal of each period of the radar is subjected to FFT processing to obtain the distance information of the target.

[0022] After the distance FFT, the obtained matrix is called a range image matrix. The echo signals of two adjacent periods have peak values at the same position, and the phase difference between the peak values is changed due to the radial velocity information of the object on the echo path. The relationship between the phase difference ω and the radial velocity v of the object is expressed as follows:

[0023]

[0024] where λ represents the wavelength of the radar signal.

[0025] Then, a Doppler FFT operation is performed, that is, the phase difference information is obtained by performing FFT on the data of the distance unit where the peak value is located on the range image.

[0026] The matrix obtained after the Doppler FFT processing is called a range-Doppler spectrum. By using constant false alarm rate (CFAR) on the range-Doppler spectrum, the range-Doppler unit where the target is located is obtained, and the distance and radial velocity information of the target are determined.

[0027] Then, the MVDR method is used to measure the angle of each distance unit of the echo, and the spatial spectrum expression of the θ0 angle is as follows:

[0028]

[0029] where a(θ) represents a direction vector, R(·) represents the autocorrelation matrix of the distance unit corresponding to the echo signal, H denotes a conjugate transpose operation, and by selecting the angle at the peak value on the selected angle spatial spectrum as the current estimated angle, the azimuth angle θ of the distance unit can be obtained.

[0030] Finally, based on the above parameters, the distance R, the radial velocity v and the angle θ of the target are obtained, and the position information [x, y] of the target is obtained through the following formula:

[0031]

[0032] Further, the step S2 is specifically as follows:

[0033] In the non-line-of-sight scene, that is, the L-shaped corner non-line-of-sight scene, it is set that the three walls in the L-shaped corner are respectively named as Wall-1, Wall-2 and Wall-3.

[0034] The target real position is set as [x r ,y r ] T , (·) T represents the transposition operation of the matrix, C represents the corner formed by Wall-1 and Wall-3, and the coordinates are [D1, D3] T , D1 and D2 respectively represent the horizontal distances of Wall-1 and Wall-2 to the radar, and D3 represents the vertical distance of Wall-3 to the radar, and the target is in the corridor formed by Wall-1 and Wall-2. A radar is placed on the other side of the corner to detect the target hidden in the non-line-of-sight area, and the radar coordinates are [0, 0] T .

[0035] The path of the radar signal reaching the target through diffraction is called the diffraction path Path-0, the path of the radar signal reaching the target after reflecting on the wall once is called the first reflection path Path-1, the path of the radar signal reaching the target after reflecting on the wall twice is called the second reflection path Path-2, and the signal passing through more than twice wall reflection is not considered.

[0036] When the target is in the non-line-of-sight environment, any combination of the above three paths is obtained in the process of the radar signal reaching the target and returning to the radar. Six kinds of combined paths are obtained, and six virtual targets are obtained through the radar signal processing procedure in step S1. The virtual targets generated by the pure diffraction path and the second reflection path and their combined paths are difficult to be observed, so they are not considered for the moment, that is, after the radar signal processing procedure in step S1, the virtual target Q 01 generated by Path-0 and Path-1 paths, the virtual target Q 11 generated by Path-1 and Path-1 paths, and the virtual target Q 12 generated by Path-1 and Path-2 paths.

[0037] After obtaining the distance and angle information of the above three virtual targets through step S1, a density-based clustering algorithm DBSCAN is used to classify the three virtual targets. The calculation formula of DBSCAN is as follows:

[0038] I = DBScan (Ψ, ε, Pts)

[0039] Where Ψ represents the set of all virtual targets, DBScan(·) represents the execution of the DBSCAN algorithm, ε represents the scan radius, Pts represents the minimum number of points within the scan radius, and I represents the category index output by the algorithm. The elements in I correspond one-to-one with the elements in Ψ.

[0040] After clustering, Ψ will be divided into 3 categories, representing Q respectively. 01 Q 11 and Q 12 The estimated maximum value for each category is the location of the virtual target. Ultimately, the location results for the three categories of virtual targets are as follows:

[0041]

[0042] Furthermore, step S3 is specifically as follows:

[0043] Given Q 12 Distance R to radar 12 Based on step S2, Q 12 The virtual target is generated by the combination of Path-1 and Path-2, so Q is known to be... 22 If Q is a virtual target generated by a combination of Path-2 and Path-2, then Q 22 Distance R to radar 22 and R 12 And Q 11 Distance R to radar 11 The relational expression is as follows:

[0044] R 11 +R 22 =2R 12

[0045] Based on the above formula, R is obtained. 22 Based on the geometric symmetry between the target and the reflection path, Q... 11 It is the mirror image of target Q about Wall-2, Q 22 Having the same y-coordinate 11 According to Q 22 The ordinate y 11 and the distance R to the radar 22 Find Q 22 coordinates [x 22 ,y 22 ] T Derivation of Q 11 =[x 11 ,y 11 ] T Q 22 =[x 22 ,y 22 ] Tand real target position Q = [x r ,y r ] T The relationship between the three is as follows:

[0046]

[0047] Considering the position information of Q 11 and Q 22 , the real target position information is obtained:

[0048]

[0049] Then, the target speed is estimated by combining the first reflection and the second reflection paths. The target speed in the horizontal and vertical directions is respectively V x and V y . The radial speed of the target on the first reflection path is V 11 , and the speed on the first-second combined path is V 12 . According to the geometric relationship between the target and the reflection path, the radial speed of the target on the second reflection path V 22 has the following relationship:

[0050] V 22 + V 11 = 2V 12

[0051] Based on the above formula, V 22 is solved.

[0052] Then, the system analysis is performed for Q 11 . The speed of Q 11 in the horizontal direction is V x1 , and the speed in the vertical direction is V y1 . The angle between the first reflection path and the radar in the vertical direction is θ1. The analysis can obtain the expression as follows:

[0053] V y1 cos(θ1) + V x1 sin(θ1) = V 11

[0054] Similarly, for Q 22 , we have:

[0055] V y2 cos(θ2) + V x2 sin(θ2) = V 22

[0056] where V x2 represents the speed of Q 22 in the horizontal direction, and V y2 represents the speed of Q22 The velocity in the vertical direction, the secondary reflection path and the angle of the radar in the vertical direction are θ2.

[0057] Then according to the geometric relationship of multipath reflection, the following expression is obtained:

[0058]

[0059] By simultaneously solving the above expressions, the following is obtained:

[0060]

[0061] Finally, V is obtained by the above formula x , V y , that is, the velocity of the target is obtained.

[0062] The application has the beneficial effects that the application discloses a non-line-of-sight target positioning and velocity estimation method based on combined path echoes, first, radar signals are processed, the distance-Doppler spectrum of a target is obtained from radar original echoes, then the distance, angle and radial velocity information of the target are obtained through the distance-Doppler spectrum, then in a non-line-of-sight scene, virtual targets are obtained in combination with known building scene layout information, the virtual targets of different paths are classified through a clustering algorithm, multi-path echo classification is completed, direct positioning results are obtained, finally, the real position of the target is obtained through fusion of multipath position information, and the real velocity of the target is obtained in combination with multipath velocity information, so that the positioning and velocity estimation of the non-line-of-sight target are realized. The method of the application can realize high-precision positioning of a non-line-of-sight area target and estimate the velocity parameter of the target, ensures the positioning and velocity estimation effect of the target in a non-line-of-sight scene, and provides strong protection for target monitoring technology in a non-line-of-sight area. BRIEF DESCRIPTION OF DRAWINGS

[0063] Figure 1 It is a flowchart of the non-line-of-sight target positioning and velocity estimation method based on combined path echoes of the application.

[0064] Figure 2 It is a simulation distance-Doppler spectrum diagram in the embodiment of the application.

[0065] Figure 3 It is a simulation result diagram of the distance-Doppler spectrum after constant false alarm detection in the embodiment of the application.

[0066] Figure 4 It is an L-shaped corner non-line-of-sight scene schematic diagram in the embodiment of the application.

[0067] Figure 5 It is a result diagram after clustering using a DBSCAN algorithm in the embodiment of the application.

[0068] Figure 6This is a two-dimensional schematic diagram of a non-line-of-sight scene simulation of an L-shaped corner in an embodiment of the present invention.

[0069] Figure 7 Q is an example of the present invention. 11 Schematic diagram of velocity decomposition.

[0070] Figure 8 This is the final positioning simulation result diagram in the embodiment of the present invention.

[0071] Figure 9 This is a positioning error curve diagram in an embodiment of the present invention.

[0072] Figure 10 This is a diagram of an actual experimental scenario in an embodiment of the present invention.

[0073] Figure 11 This is a measured distance-Doppler spectrum in an embodiment of the present invention.

[0074] Figure 12 This is a diagram showing the actual measurement and positioning results in an embodiment of the present invention. Detailed Implementation

[0075] The method of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0076] This example is a MATLAB simulation example, such as... Figure 1 The flowchart of a non-line-of-sight target localization and velocity estimation method based on combined path echoes of the present invention is shown below. The specific steps are as follows:

[0077] S1. Process the radar signal to obtain the target's range-Doppler spectrum from the original radar echo, and then obtain the target's range, angle, and radial velocity information through the range-Doppler spectrum;

[0078] S2. In non-line-of-sight scenarios, based on the known building scene layout information, virtual targets are obtained from step S1. Virtual targets on different paths are classified using a clustering algorithm to complete multi-path echo classification and obtain direct positioning results.

[0079] S3. Based on step S2, the true position of the target is obtained by fusing multipath position information, and the true velocity of the target is obtained by combining multipath velocity information, thereby realizing the localization and velocity estimation of non-line-of-sight targets.

[0080] In this embodiment, step S1 is specifically as follows:

[0081] Both the simulation and actual test scenarios use millimeter-wave radar, and the radar transmits a linear frequency modulated (LFM) signal. The radar's carrier frequency is set to f. c The signal bandwidth is B, and the signal period is T. c The signal amplitude is AT The phase of the signal is The signal transmission time is t, and the transmission signal x T (t) is expressed as follows:

[0082]

[0083] The transmission signal encounters a target and reflects a return signal, which is received by the receiving antenna. The received signal x R (t) is expressed as follows:

[0084]

[0085] where A R represents the amplitude of the received signal, and τ represents the time delay of the signal, i.e., the time for the signal to travel from the radar to the target and then return to the radar. In a multipath environment, the expression is as follows:

[0086]

[0087] where R represents the distance from the radar to the target, and c represents the speed of light.

[0088] After the receiving antenna receives the return signal reflected by the target, the return signal is mixed with the transmission signal to obtain a difference frequency signal y(t), also known as an intermediate frequency signal. The phase of the intermediate frequency signal is the phase difference between the return signal and the transmission signal, and the expression is as follows:

[0089]

[0090] where α represents the signal amplitude coefficient, represents the phase of the intermediate frequency signal, represents the high-order term about t generated in the mixed signal, which can be ignored, and f b represents the frequency of the intermediate frequency signal, and the specific expression is as follows:

[0091]

[0092] After obtaining the intermediate frequency signal, the intermediate frequency signal is subjected to Fast Fourier Transform (FFT) processing, i.e., the frequency f b, and then the distance R of the target to the radar is obtained. In actual measurement, the original echo signal collected by the radar is the intermediate frequency signal after mixing, and thus the distance information of the target can be obtained by directly performing FFT processing on the original echo signal of each cycle of the radar, which is called distance FFT, and the obtained matrix is called a range image matrix. Since the time interval between two adjacent cycles during the operation of the radar is very short, the moving speed of the object can be considered unchanged in this time interval, and thus the echo signals of two adjacent cycles will have peak values at the same position after distance FFT, but the phase difference between the peak values will change due to the radial speed information of the object on the echo path. The relationship between the phase difference ω and the radial speed v of the object is expressed as follows:

[0093]

[0094] wherein λ represents the wavelength of the radar signal.

[0095] The phase difference information can be obtained by performing FFT on the data of the distance unit where the peak value is located on the range image, which is called Doppler FFT. The matrix obtained after Doppler FFT is called a range-Doppler spectrum, and the distance-Doppler unit where the target is located can be obtained by using constant false alarm detection (CFAR) on the range-Doppler spectrum, and then the distance and radial speed information of the target can be determined.

[0096] Then, the MVDR (Minimum Variance Distortionless Response) method is used to measure the angle of each distance unit of the echo, and the spatial spectrum expression of the θ0 angle is as follows:

[0097]

[0098] wherein a(θ) represents a direction vector, represents the autocorrelation matrix of the echo signal corresponding to the distance unit, (·) H represents a conjugate transpose operation, and the angle at the peak value on the selected angle spatial spectrum is taken as the current estimated angle, so as to obtain the azimuth angle θ of the distance unit.

[0099] The radar signal processing part of the embodiment ends, and based on the above parameters, the distance R, radial speed v and angle θ of the target can be obtained, and the position information [x, y] of the target can be obtained through the following formula:

[0100]

[0101] In the ordinary line-of-sight scenario, the distance, angle and radial velocity information of the target has been obtained, but in the non-line-of-sight scenario, due to the reflection and propagation of electromagnetic waves on the wall and other buildings, the information obtained by the above signal processing procedure is the distance, angle and radial velocity information of the virtual target from multiple paths, which is different from the information of the real target, and further processing is needed to obtain the information of the real target.

[0102] In this embodiment, after radar signal processing on the simulated radar raw echo, the range-Doppler spectrum is obtained as shown in Figure 2 , and the constant false alarm detection is performed on the range-Doppler graph to obtain 32 detection results, and the information such as the distance unit, Doppler unit and signal strength of the unit where they are located is recorded as shown in Figure 3 .

[0103] In this embodiment, the step S2 is specifically as follows:

[0104] The radar signal processing procedure in step S1 obtains target information that can be directly used for positioning to obtain the target position in the general line-of-sight case (i.e. the target is in the line-of-sight area of the radar, and the radar signal can directly reach the target), but when the target is in the non-line-of-sight area, the obtained target information is different from the real target information, and at this time, to position the target, the scene layout information needs to be further processed.

[0105] As shown in Figure 4 , the L-shaped corner non-line-of-sight scenario (effective multipath propagation model) schematic diagram, the general scenario applicable to this embodiment is the L-shaped corner non-line-of-sight scenario, and the three walls in the L-shaped corner are named as Wall-1, Wall-2 and Wall-3.

[0106] The real position of the target is set as [x r ,y r ] T , (·) T represents the transpose operation of the matrix, C represents the corner formed by Wall-1 and Wall-3, and the coordinates are [D1, D3] T , D1 and D2 respectively represent the horizontal distances of Wall-1 and Wall-2 to the radar, and D3 represents the vertical distance of Wall-3 to the radar, the target is in the corridor formed by Wall-1 and Wall-2, and a radar is placed on the other side of the corner to detect the target hidden in the non-line-of-sight area, and the radar coordinates are [0, 0] T .

[0107] According to the multi-path propagation characteristics of electromagnetic waves, the radar signal will reach the target through the mirror reflection of the wall. The path through which the radar signal reaches the target by diffraction is called the diffraction path Path-0, the path through which the radar signal reaches the target after being reflected once on the wall is called the first reflection path Path-1, and the path through which the radar signal reaches the target after being reflected twice on the wall is called the second reflection path Path-2. Since each reflection of the signal at the wall causes attenuation, signals that have been reflected more than twice on the wall are not considered.

[0108] According to the propagation characteristics of electromagnetic waves, when the signal echo reaches the target along Path-1, the return path can be Path-1, Path-0, or Path-2. The same applies to the signal echoes that reach the target along the other two paths. The return path exists for all three paths, i.e., when the target is in a non-line-of-sight environment, the three paths can be combined in any way during the process of the radar signal reaching the target and returning to the radar. As a result, there are six combined paths, and six virtual targets are generated. In theory, the virtual targets obtained by the radar signal processing flow in step S1 have six types, but since the diffraction and second reflection of the signal are greatly attenuated, it can be considered that the virtual targets generated by the pure diffraction path and the second reflection path and their combined paths are difficult to observe, and can be temporarily ignored. That is, in the actual situation, after the radar signal processing flow in step S1, the virtual targets Q 01 generated by Path-0 and Path-1, the virtual targets Q 11 generated by Path-1 and Path-1, and the virtual targets Q 12 generated by Path-1 and Path-2 are obtained.

[0109] The calculation formula of DBSCAN is as follows:

[0110] I = DBScan(Ψ, ε, Pts)

[0111] wherein Ψ represents the set of all virtual targets, DBScan(·) represents the execution of the DBSCAN algorithm, ε represents the scanning radius, Pts represents the minimum number of points within the scanning radius, and I represents the class index output by the algorithm, the elements of which correspond one-to-one to the elements in Ψ.

[0112] After clustering, Ψ is divided into three categories, representing Q 01 , Q 11 , and Q12 The clustering results are as follows Figure 5 As shown. The estimated maximum value for each category is the location of the virtual target. Finally, the direct location results for the three categories of virtual targets are:

[0113]

[0114] In this embodiment, a two-dimensional simulation diagram of the non-line-of-sight scene at the L-shaped corner is shown below. Figure 6 As shown, the radar used in this embodiment, with one transmitter and four receivers, is located at the zero point of the coordinate system. The transmitting and receiving antennas are arranged horizontally, spaced half a wavelength apart. The radar's transmitted signal is a linear frequency modulated signal with a carrier frequency of 77 GHz and a bandwidth of 2.48 GHz. Wall-1 is placed vertically with an abscissa of 0.25 m, Wall-2 is placed horizontally with an ordinate of 0.23 m, and Wall-3 is placed vertically with an abscissa of 2.95 m. The target is located at (2.05, 2.5) m, and the target's horizontal and vertical velocities are both set to 2 m / s. The noise in the two-dimensional space is uniform noise, satisfying a zero-mean complex Gaussian distribution. The false alarm probability of constant false alarm detection is set to 1%, and 100 Monte Carlo simulations are performed at each signal-to-noise ratio. The scanning radius of the DBSCAN algorithm is set to the radar's range resolution, and the number of points within the scanning area is set to 5.

[0115] In this embodiment, step S3 is specifically as follows:

[0116] Given Q 12 Distance R to radar 12 Based on step S2, Q 12 The virtual target is generated by the combination of Path-1 and Path-2, so Q is known to be... 22 If Q is a virtual target generated by a combination of Path-2 and Path-2, then Q 22 Distance R to radar 22 and R 12 And Q 11 Distance R to radar 11 The relational expression is as follows:

[0117] R 11 +R 22 =2R 12

[0118] Based on the above formula, R is obtained. 22 Based on the geometric symmetry between the target and the reflection path, we can know that Q 11 It is the mirror image of target Q about Wall-2, Q 22 Having the same y-coordinate 11 According to Q 22 The ordinate y 11 and the distance R to the radar22 Q can then be obtained 22 coordinates [x 22 ,y 22 ] T Then derive Q 11 =[x 11 ,y 11 ] T Q 22 =[x 22 ,y 22 ] T And the true target position Q = [x r ,y r ] T The following relationship exists between these three:

[0119]

[0120] Taking Q into consideration 11 and Q 22 By obtaining the location information, we can obtain the target's true location information:

[0121]

[0122] Then, the target velocity is estimated by combining the paths of the first and second reflections, with the target's velocities in the horizontal and vertical directions set as V, respectively. x V y The radial velocity V along the target's first reflection path can be measured using radar. 11 And the velocity V on the first and second combined path 12 The radial velocity V on the secondary reflection path of the target can be determined from the geometric relationship between the target and the reflection path. 22 The following relationship exists:

[0123] V 22 +V 11 =2V 12

[0124] Based on the above formula, V 22 Solve the problem.

[0125] like Figure 7 As shown, now regarding Q 11 Perform a coherence analysis and define Q. 11 The velocity in the horizontal direction is V x1 The vertical velocity is V y1 The angle between the primary reflection path and the radar in the vertical direction is θ1. Analysis yields the following expression:

[0126] V y1 cos(θ1)+V x1 sin(θ1)=V11

[0127] Q 22 has:

[0128] V y2 cos(θ2)+V x2 sin(θ2)=V 22

[0129] wherein, V x2 represents Q 22 the speed in the horizontal direction, V y2 represents Q 22 the speed in the vertical direction, the angle between the secondary reflection path and the radar in the vertical direction is θ2.

[0130] Then according to the geometric relationship of multipath reflection, the following expression is obtained:

[0131]

[0132] By solving the above expressions, we get:

[0133]

[0134] Finally, V x , V y , that is, the speed of the target is obtained.

[0135] In this embodiment, after step S2, Q 11 position is [3.9808, 2.5783], Q 12 to the radar distance R 12 is 6.3037m, substituted into the formula to calculate Q 22 to the radar distance R 22 is 7.8646m, combined with Q 22 and Q 11 have the same longitudinal coordinate, the final position of Q 22 is [7.4301, 2.5783], and then substituted into the formula, the real position of the target based on the combined path is estimated to be [1.9746, 2.5783], and the virtual target Q 11 formed by the existing method based on the first reflection path is estimated to be [1.9192, 2.5783], it can be seen that in this embodiment, the positioning result of the method of the present application is better than that of the existing method.

[0136] In this embodiment, the virtual target position and radial velocity obtained by step S1 and step S2 can obtain the angle θ1 of Q 11 0.9961 rad, and the speed V 11-0.5491 m / s, Q 12 the velocity V 12 0.9609 m / s, Q 22 the angle θ1 is 0.1.2368 rad, by substituting V 11 and V 12 into the formula, Q 22 the velocity V 22 2.4709 m / s, combining the above information, the equation in step S3 is solved, and the final target velocity estimation result V x 1.9717 m / s, V y 1.9312 m / s, and the final simulation positioning and velocity estimation result is shown in Figure 8 .

[0137] In addition, in order to verify the effectiveness of the method, the embodiment also carries out Monte Carlo simulation under different signal-to-noise ratios. After simulation, the final positioning error curve is shown in Figure 9 . From the positioning error curve, it can be seen that under the condition of lower signal-to-noise ratio from -20 dB to -13 dB, the one-time path positioning result of the existing method is more accurate, and under the condition of higher signal-to-noise ratio from -13 dB to 5 dB, the combined path positioning result of the method is more accurate, which proves that the positioning result of the method under the condition of high signal-to-noise ratio is superior to the existing positioning method.

[0138] The embodiment also carries out a real test experiment. The experimental scene is shown in Figure 10 . The scene arrangement and radar parameters are the same as the simulation in the above embodiment. Since it is difficult to grasp the human target motion in the real test, the human target is set as a motion mode of shaking in place, so as to verify the accuracy of the velocity estimation. The final measured range-Doppler spectrum is shown in Figure 11 . From the spectrum, the peak values of the three types of virtual targets can be obviously observed, which is consistent with the theoretical simulation result. The final measured positioning result is shown in Figure 12 .

[0139] In summary, from the simulation and test results, it can be seen that compared with the existing non-line-of-sight positioning method using only one reflection path, the positioning effect of the method is superior to the existing positioning method, and the target velocity estimation can be realized, which verifies the correctness and effectiveness of the method. The method is more accurate for non-line-of-sight target positioning, and can estimate the velocity parameter of the target, ensuring the target positioning and velocity estimation effect in the non-line-of-sight environment, and providing a strong guarantee for the target monitoring technology in the non-line-of-sight area.

[0140] Those skilled in the art will appreciate that the embodiments described herein are presented for purposes of illustration and understanding of the principles of the application and should not be construed as limiting the scope of the application to such specifically enumerated embodiments. Various modifications and changes can be made thereto by those skilled in the art without departing from the spirit and principles of the application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the application should be included in the scope of the claims of the application.

Claims

1. A non-line-of-sight target localization and velocity estimation method based on combined path echoes, the specific steps of which are as follows: S1. Process the radar signal to obtain the target's range-Doppler spectrum from the original radar echo, and then obtain the target's range, angle, and radial velocity information through the range-Doppler spectrum; S2. In non-line-of-sight scenarios, based on the known building scene layout information, virtual targets are obtained from step S1. Virtual targets on different paths are classified using a clustering algorithm to complete multi-path echo classification and obtain direct positioning results. S3. Based on step S2, the true position of the target is obtained by fusing multipath position information, and the true velocity of the target is obtained by combining multipath velocity information, thereby realizing the localization and velocity estimation of non-line-of-sight targets.

2. The non-line-of-sight target localization and velocity estimation method based on combined path echoes according to claim 1, characterized in that, The specific steps of S1 are as follows: Both the simulation and actual test scenarios use millimeter-wave radar, and the radar transmits a linear frequency modulated (LFM) signal. The radar's carrier frequency is set to f. c The signal bandwidth is B, and the signal period is T. c The signal amplitude is A T The signal phase is If the signal transmission time is t, then the radar's transmitted signal x T The expression for (t) is as follows: The transmitted signal is reflected back after encountering the target and received by the receiving antenna; the received signal is x. R The expression for (t) is as follows: Among them, A R The amplitude of the received signal is represented by τ, and the signal delay is represented by τ, which is the time it takes for the signal to travel from the radar to the target and back to the radar. In a multipath environment, the expression is as follows: Where R represents the distance from the radar to the target, and c represents the speed of light; After the receiving antenna receives the echo signal reflected back from the target, it mixes it with the transmitted signal to obtain the difference frequency signal y(t), also known as the intermediate frequency signal. The phase of the intermediate frequency signal is the phase difference between the echo signal and the transmitted signal, and its expression is as follows: Where α represents the signal amplitude coefficient, Indicates the phase of the intermediate frequency signal. f represents the higher-order terms with respect to t generated in the mixing signal, which can be ignored. b The frequency of the intermediate frequency signal is represented by the following expression: After obtaining the intermediate frequency (IF) signal, a Fast Fourier Transform (FFT) is performed on the IF signal to obtain the frequency f of the IF signal. b This allows us to obtain the distance R from the target to the radar. Since the original echo signal collected by the radar is the intermediate frequency signal after mixing, we perform a range FFT operation, which means directly performing FFT processing on the original echo signal of each radar cycle to obtain the target's distance information. After the range FFT, the obtained matrix is ​​called the range image matrix. The echo signals of two adjacent periods have peaks at the same location. The phase difference between the peaks will change due to the radial velocity information of the object on the echo path. The relationship between the phase difference ω and the radial velocity v of the object is expressed as follows: Where λ represents the wavelength of the radar signal; Then, a Doppler FFT operation is performed, that is, the phase difference information is obtained by performing FFT on the data of the range cell where the peak is located on the range image; The matrix obtained after Doppler FFT processing is called the range-Doppler spectrum. By applying constant false alarm rate (CFAR) detection to the range-Doppler spectrum, the range-Doppler cell where the target is located can be obtained, and the range and radial velocity information of the target can be determined. Then, the MVDR method was used to measure the angle of each distance cell in the echo, and the spatial spectrum expression of the angle θ0 is as follows: Where a(θ) represents the direction vector. Let represent the autocorrelation matrix of the range cell corresponding to the echo signal, (·). H This indicates the conjugate transpose operation, which uses the angle at the peak of the selected angle space spectrum as the current estimated angle to obtain the azimuth angle θ of the distance cell; Finally, based on the above parameters, the target's distance R, radial velocity v, and angle θ are obtained, and the target's position information [x,y] is obtained using the following formula:

3. The non-line-of-sight target localization and velocity estimation method based on combined path echoes according to claim 1, characterized in that, Step S2 is as follows: In non-line-of-sight scenes, i.e., L-shaped corner non-line-of-sight scenes, the three walls in the L-shaped corner are named Wall-1, Wall-2 and Wall-3 respectively. Set the target's actual location as [xr, yr] T ,(·) T Let C represent the transpose of the matrix, where C represents the corner formed by Wall-1 and Wall-3, with coordinates [D1, D3]. T D1 and D2 represent the horizontal distances from Wall-1 and Wall-2 to the radar, respectively, and D3 represents the vertical distance from Wall-3 to the radar. The target is located in the corridor formed by Wall-1 and Wall-2. A radar is placed on the other side of the corner to detect targets hidden in the non-line-of-sight area, with radar coordinates [0,0]. T ; The path of a radar signal reaching the target through diffraction is called the diffraction path Path-0. The path of a radar signal reaching the target after being reflected once by a wall is called the first reflection path Path-1. The path of a radar signal reaching the target after being reflected twice by a wall is called the second reflection path Path-2. Signals that have been reflected more than twice by walls are not considered. When the target is in a non-line-of-sight environment, during the process of the radar signal reaching the target and returning to the radar, the above three paths can be arbitrarily combined to obtain six combined paths. After the radar signal processing flow in step S1, six virtual targets are obtained. Virtual targets generated by pure diffraction paths, secondary reflection paths, and their combinations are difficult to observe, so they are temporarily ignored. That is, after the radar signal processing flow in step S1, the virtual target Q generated by Path-0 and Path-1 paths is obtained. 01 The virtual target Q generated by Path-1 and Path-1 path 11 The virtual target Q generated by Path-1 and Path-2 12 ; After obtaining the distance and angle information of the three virtual targets in step S1, the density-based clustering algorithm DBSCAN is used to classify these three virtual targets; the calculation formula of DBSCAN is as follows: I = DBScan(Ψ,ε,Pts) Where Ψ represents the set of all virtual targets, DBScan(·) means to execute the DBSCAN algorithm, ε represents the scanning radius, Pts represents the minimum number of points within the scanning radius, and I represents the category index output by the algorithm. The elements in I correspond one-to-one with the elements in Ψ. After clustering, Ψ will be divided into 3 categories, representing Q respectively. 01 Q 11 and Q 12 The estimated maximum value for each category is the location of the virtual target. Finally, the location results for the three categories of virtual targets are as follows:

4. The non-line-of-sight target localization and velocity estimation method based on combined path echoes according to claim 1, characterized in that, Step S3 is as follows: Given Q 12 Distance R to radar 12 Based on step S2, Q 12 The virtual target is generated by the combination of Path-1 and Path-2, so Q is known to be... 22 If Q is a virtual target generated by a combination of Path-2 and Path-2, then Q 22 Distance R to radar 22 and R 12 And Q 11 Distance R to radar 11 The relational expression is as follows: R 11 +R 22 =2R 12 Based on the above formula, R is obtained. 22 Based on the geometric symmetry between the target and the reflection path, Q 11 It is the mirror image of target Q about Wall-2, Q 22 Having the same y-coordinate 11 According to Q 22 The ordinate y 11 and the distance R to the radar 22 Find Q 22 coordinates [x 22 ,y 22 ] T Derivation of Q 11 =[x 11 ,y 11 ] T Q 22 =[x 22 ,y 22 ] T And the true target position Q = [x r ,y r ] T The following relationship exists between these three: Taking Q into consideration 11 and Q 22 From the location information, we obtain the target's true location information: Then, the target velocity is estimated by combining the paths of the first and second reflections, with the target's velocities in the horizontal and vertical directions set as V, respectively. x V y The radial velocity V along the target's first reflection path can be measured using radar. 11 And the velocity V on the first and second combined path 12 Based on the geometric relationship between the target and the reflection path, the radial velocity V on the target's secondary reflection path can be determined. 22 The following relationship exists: V 22 +V 11 =2V 12 Based on the above formula, V 22 Solve the problem; Then regarding Q 11 Perform a coherence analysis and define Q. 11 The velocity in the horizontal direction is V x1 The vertical velocity is V y1 The angle between the primary reflection path and the radar in the vertical direction is θ1; the following expression can be obtained through analysis: V y1 cos(θ1)+V x1 sin(θ1)=V 11 Similarly, for Q 22 have: V y2 cos(θ2)+V x2 sin(θ2)=V 22 Among them, V x2 Q represents 22 The velocity in the horizontal direction, V y2 Q represents 22 The vertical velocity, the secondary reflection path, and the angle between the radar in the vertical direction are θ2; Then, based on the geometric relationship of multipath reflection, we have the following expression: By combining the above expressions, we get: Finally, V is obtained using the above formula. x V y That is, the speed of the target was obtained.

Images