A two-stage target localization method for long L-shaped building layout

Abstract

The present invention discloses a two-stage target positioning method under a long L-shaped building layout, which is applied to the field of target positioning technology in complex urban environments. In order to solve the problem that the existing technology can only effectively detect targets in the first-order reflection area, the present invention first divides the long L-shaped scene into multiple reflection areas according to the ray tracing model, and analyzes the electromagnetic propagation model therein respectively; then, based on the fact that diffraction paths exist in all non-line-of-sight areas, the area where the target may exist is delineated using the time delay corresponding to the diffraction path; secondly, the area is divided into different grids according to the radar range resolution, and the corresponding theoretical time delay is calculated according to the established multipath model; finally, the accurate target position is obtained by matching the grid theoretical multipath delay with the received echo multipath delay. The method of the present invention can realize non-line-of-sight target positioning in a large depth scene, and reduces the complexity of the calculation while ensuring the positioning accuracy.

Description

Technical Field

[0001] The present invention belongs to the technical field of multipath radar target positioning, and in particular to a single target positioning technology when the target spans multiple non-line-of-sight areas and the multipath echo types are complex. Background Art

[0002] Detecting obscured targets in complex urban environments holds important research value in various fields, including intelligent driving. In these situations, targets are often obscured by multiple obstacles, preventing electromagnetic waves from propagating along the line-of-sight (LOS) path to the target. Therefore, non-LOS (NLOS) detection methods based on diffraction from wall corners and reflection from building surfaces are used to locate hidden targets.

[0003] In general, existing non-line-of-sight target localization methods can be roughly divided into two categories: one based on multipath identification and the other based on multipath accumulation. The core of the first category of methods lies in determining the theoretical multipath corresponding to the time of arrival (ToA) in the echo. The target position can then be derived using the geometric relationship of the multipath. The Ilmanau University of Technology in Germany obtained the one-dimensional position of the target based on the geometric symmetry relationship of a single mirror reflection (R. Zetik, M. Roding, and R. Thoma, "UWB localization of moving targets in shadowed regions," Proc. 6th Eur. Conf. Antennas Propag., Prague, Czech Republic, pp. 1729-1732, June 2012). Subsequently, the two-dimensional target position of the non-line-of-sight target was obtained by intersecting the diffraction path and the multipath trajectory corresponding to the first-order reflection (R. Zetik, M. Eschrich, S. Jovanoska, and RS Thoma, “Looking behind a corner using multipath-exploiting UWB radar,” IEEE Trans. Aerosp. Electron. Syst., vol. 51, no. 3, pp. 1916-1926, July 2015). However, the above two methods fail to fully utilize multipath and cannot effectively deal with the situation where multipath is missing. To solve this problem, the University of Electronic Science and Technology of China implemented multipath identification through the ToA correlation method and then obtained the target position based on the ellipse intersection positioning method (X. Yang, S. Fan, S. Guo, S. Li, G. Cui and W. Zhang, “NLOS Target Localization Behind an L-Shaped Corner With an L-Band UWB Radar,” IEEE Access, vol. 8, pp. 31270-31286, 2020). However, this type of method relies on the accuracy of the extracted multipath ToA, and the positioning accuracy is low when the measurement error is large.

[0004] The second category of methods uses imaging to accumulate multipath energy without identifying the extracted ToA. In the paper (S. Li, G. Cui, S. Guo, H. Li, L. Kong and X. Yang, “NLOS Targets Imaging with UWB Radar,” Int. Conf. Control Automat. Inf. Sci., Chengdu, China, pp. 1-5, Oct. 2019), the University of Electronic Science and Technology of China used backprojection to obtain images of different multipaths and then multiply them to produce a radar image containing only the true target locations. Furthermore, considering that the first-order ghosts and the true targets in the backprojected image are symmetrical, the paper (S. Li et al., “Multiple targets localization behind L-shaped corner via UWBradar,” IEEE Trans. Veh. Technol., vol. 70, no. 4, pp. 3087–3100, Apr. 2021) used spatial position matching of multipath ghosts to achieve the location determination of multiple non-line-of-sight targets. However, the back-projection imaging method will cause the clutter energy to accumulate, resulting in the appearance of false targets. The National University of Defense Technology proposed a grid matching method, matching the theoretical multipath ToA of each grid with the ToA extracted from the echo to determine the location of the real target (H.Du, C.Fan, Z.Chen, C.Cao, X.Huang, "NLOS Target Localization with an L-Band UWB Radar via GridMatching," Prog.In Electromagn.Res., vol.97, pp.45-56, 2020). However, the above research only considered the case where the obscured target is located within the first-order reflection area, which undoubtedly limits the detectable range of these methods. Therefore, it is of great research significance to study a non-line-of-sight target positioning method in a long L-shaped scene, which can solve the positioning of the target when it spans multiple non-line-of-sight areas. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes a two-stage target positioning method under a long L-shaped building layout, which can effectively and accurately estimate the target position across multiple non-line-of-sight areas.

[0006] The technical solution adopted by the present invention is: a two-stage target positioning method in a long L-shaped building layout, the application scenario includes: wall 1, wall surface 2, wall surface 3, a single-transmitter single-receiver radar, and a hidden target. Wall 1 and wall surface 2 form an L-shaped wall, and wall surface 2 and wall surface 3 are parallel. The single-transmitter single-receiver radar is set in front of wall 1, and the hidden target is located in the non-line-of-sight area between wall surfaces 2 and wall 3. The positioning method specifically includes the following steps:

[0007] S1. Divide the non-line-of-sight area into different reflection areas based on ray tracing, and then analyze the multipath propagation model in different reflection areas respectively;

[0008] S2. Determine the target potential area based on the diffraction delay according to the constructed multipath propagation model;

[0009] S3. Divide the target potential area into grids according to the radar range resolution, and calculate the corresponding theoretical multipath delay;

[0010] S4. Match the arrival time of the received echo with the theoretical multipath delay of the grid to obtain the true position of the target.

[0011] Beneficial effects of the present invention: The method of the present invention first divides a long L-shaped scene into multiple reflection areas based on a ray tracing model and analyzes the electromagnetic propagation models therein respectively. Then, based on the fact that diffraction paths exist in all non-line-of-sight areas, the time delays corresponding to the diffraction paths are used to delineate the area where the target may be located. Secondly, the area is divided into different grids based on the radar range resolution, and the corresponding theoretical time delays are calculated based on the established multipath model. Finally, the accurate target position is obtained by matching the grid theoretical multipath delay with the received echo multipath delay. The method of the present invention has the following advantages:

[0012] 1. The proposed multipath model can effectively extend the detection range of traditional non-line-of-sight target positioning algorithms;

[0013] 2. The positioning algorithm has low computational complexity and high positioning accuracy;

[0014] 3. Effectively solves the positioning problem of non-line-of-sight targets across multiple areas;

[0015] 4. The present invention can be applied to fields such as disaster relief and medical monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Modeling diagram for the long L-shaped building scene.

[0017] Figure 2 Schematic diagram of the multipath propagation model in the fourth-order reflection area.

[0018] Figure 3 for Figure 2Single-cycle target echo in the simulation scenario shown.

[0019] Figure 4 Schematic diagram of the simulation scene.

[0020] Figure 5 for Figure 4 Positioning results obtained from the simulation scenario shown;

[0021] in, Figure 5 (a) and Figure 5 (b) Positioning results of the 45th and 96th cycles respectively; Figure 5 (c) is the positioning result of all 100 cycles; Figure 5 (d) Comparison of the positioning error of the algorithm of the present invention and the positioning error obtained by the grid matching method. DETAILED DESCRIPTION

[0022] To facilitate those skilled in the art to understand the technical content of the present invention, the present invention is further explained below with reference to the accompanying drawings.

[0023] A two-stage target positioning method for a long L-shaped building layout of the present invention comprises the following steps:

[0024] Step 1: Divide the detection scene into non-line-of-sight areas:

[0025] A long L-shaped building scene consists of Building 1 and Building 2, such as Figure 1 As shown. The corner point formed by wall 1 and wall 2 is C = [x c ,y c ] T , where the superscript T indicates transposition. The horizontal coordinates of wall 2 and wall 3 are x w2 , x w3 In order to measure the hidden target P on one side of the wall 2, p ,y p ] T , a single input single output (SISO) radar R = [x r ,y r ] T Placed on the side of wall 1. At this point, the target is invisible to the radar. The superscript T indicates transposition.

[0026] Typically, ray tracing is used to describe and analyze the propagation of electromagnetic waves. Therefore, it is not difficult to find that the straight line RM1 connecting the radar and the wall corner is the boundary of the first-order reflection. When the electromagnetic wave propagates along this path, a first-order reflection occurs at M1 on wall 3, followed by a second-order reflection at M2 on wall 2. Similarly, third-order and fourth-order reflections occur at M3 and M4. Therefore, the non-line-of-sight area is divided into different triangular areas, which are named first-order reflection area, second-order reflection area, etc., as shown in Figure 2. Figure 1 When the shielding target is located in different reflection areas, the multipath signals are significantly different. Therefore, the present invention further refines the multipath propagation model in the non-line-of-sight area, as follows:

[0027] First-order reflection area: This is a typical non-line-of-sight area, and most existing algorithms are designed for this type of scene. Figure 1 As shown, target detection can be achieved through diffraction and reflection. Ignoring the high-order reflection paths with large electromagnetic wave attenuation, only the four one-way propagation paths of electromagnetic waves are considered, namely the diffraction path Path-0, the first-order reflection path Path-1, the second-order reflection path Path-3, and the third-order reflection path Path-4. To distinguish them from subsequent paths, the above three reflection paths are referred to as simple reflection paths;

[0028] Second-order reflection area: In this area, due to the existence of the reflection boundary, target detection cannot be achieved through first-order reflection, but only through diffraction and higher-order (≥2) reflection. However, this does not mean that the first reflection disappears in this area. The electromagnetic wave first diffracts at the corner of the wall, and then experiences first-order reflection on the wall 3 to propagate to the target. Its existence form is as follows Figure 2 As shown. The above first-order reflection path is called the combined reflection path and is still recorded as Path-1. Path-2 and Path-3 in this area still exist in the form of simple reflection paths;

[0029] Third-order reflection area: In this area, in addition to the first-order reflection, the second-order reflection also exists in the form of combined reflection, while the third-order reflection is still a simple reflection;

[0030] Fourth-order reflection area: such as Figure 2 As shown, the first-order reflection path, the second-order reflection path and the third-order reflection path in this area are all combined reflection paths.

[0031] Assuming that all reflections are specular, the radar virtual mirror positions associated with Path-1, Path-2, and Path-3 can be calculated using geometric relationships as follows:

[0032]

[0033] Therefore, by mirroring the radar position, the length of the simple reflection path Path-m can be expressed as ||R m Similarly, for the combined reflection path, the corner point can be mirrored:

[0034]

[0035] Combining the multipath propagation model analyzed above, the propagation delays of the four one-way paths are obtained as follows:

[0036]

[0037] Here, c represents the propagation speed of electromagnetic waves in air, and ||·|| represents the second norm. For convenience, the corner point C is denoted as C0.

[0038] Since both the transmitted signal and the received signal can propagate along the four paths mentioned above, there are a total of ten different two-way transmission path combinations. These combined paths are placed in the vector P (k) , recorded as:

[0039]

[0040] in, represents the two-way propagation path consisting of the transmit path Path-m and the receive path Path-n in the k-order reflection region. The corresponding propagation delay is:

[0041]

[0042] Step 2: Identify potential target areas:

[0043] 2.1 Use the Moving Target Indicator (MTI) technology to eliminate the influence of static background echo and antenna coupling in the received echo signal.

[0044] 2.2 Use the Constant False Alarm Rate (CFAR) detector to extract the ToA of different paths in the echo and store them in vector T in ascending order:

[0045] T=[τ1,τ2,…,τ L ] T ,τ1≤τ2≤…≤τ L (6)2.3 Through the multi-level propagation model established above, it is not difficult to find that diffraction exists in the reflection area of ​​each order, so it can be used to preliminarily determine the target area. Specifically, based on the fact that the two-way diffraction path distance is the shortest, τ1 is extracted and considered to be the two-way diffraction propagation delay. Since the distance between the wall corner and the target has the following relationship:

[0046] ||C0P||=(τ1c-2||RC0||) / 2=r p (7)

[0047] Obviously, the above formula indicates that the target is located at the corner point C as the center, r p The part of the non-line-of-sight region is represented by an arc. Theoretically, the obstructing target is located on this arc, but considering the existence of measurement errors, the extracted τ1 is not completely equivalent to the theoretical value. Therefore, the arc where the target is located is expanded into an area Ω:

[0048]

[0049] Among them, △R u represents the radar measurement error, r max and r min Represent the upper and lower bounds of the target potential area respectively. u It is generally set to be greater than three times the radar range resolution.

[0050] Step 3: Target potential area mesh division:

[0051] The target potential area Ω is divided into S grids, whose side length is set to be smaller than the radar range resolution d. The center coordinate of each grid is G s =[x s ,y s ] T ,s∈{1,2,…,S}.

[0052] Due to the scene layout prior, the boundaries of each order of reflection can be Figure 1 The model shown is easily derived. Therefore, knowing the grid G s After the reflection area is located, the corresponding two-way theoretical transmission delay is calculated and stored in the vector T s middle:

[0053]

[0054] in, Represents the grid G s The round-trip transmission delay is calculated as shown in formula (5), where the superscript k indicating the area where the grid is located is replaced by the index value s of the grid.

[0055] Step 4: Get the real target position:

[0056] 4.1 For evaluation T and grid T s Similarity, construct two matching factors N s and E s , named as matching factor and error factor. N sIndicates T s The number of ToAs that successfully match in T, E s Represents the sum of the corresponding matching errors, and its expression is:

[0057]

[0058] in,

[0059]

[0060] Here, δ represents an empirical threshold with a small value. In the specific implementation, δ is set to 0.1m / c (c is the speed of light), and 0.1m is 0.5 times the target size.

[0061] 4.2 Traverse all grids in the target potential area and store all the matching factors and error factors obtained in two vectors:

[0062]

[0063] Formulate the best matching rule to determine the grid where the real target is located, that is: N s The largest grid corresponds to E s The smallest one.

[0064] The following describes a specific implementation of the present invention based on a gprMax simulation example.

[0065] Step 1: Divide the detection scene into non-line-of-sight areas:

[0066] A target moves in a long L-shaped scene composed of multiple reflective areas, such as Figure 3 As shown, the scene is constructed and simulated using gprMax. The wall is assumed to be uniform and its relative dielectric constant is ε = 7. The coordinates of the corner are [1.5m, 1.5m] T , the horizontal coordinate of wall 3 is 5m and the length is 13m. According to the scene information, a cylinder with a radius of 0.2m and a relative dielectric constant ε = 7 is used to simulate the human target. The target is located at point P = [2.0m, 3.0m] in the first-order reflection area. T Start and move in a straight line to point Q = [4.0m, 12.0m] in the fifth-order reflection area. T The radar position is [0.5m, 0.9m] T The transmitted waveform is a Ricker wavelet with a center frequency of 1.5 GHz.

[0067] According to the relationship between radar coordinates and scene geometry, the virtual radar coordinates corresponding to the simple path are obtained as follows:

[0068]

[0069] The virtual corner position corresponding to the combined path is:

[0070]

[0071] In order to verify the correctness of the above model, the electromagnetic simulation software gprMax was used to obtain the received echo signal of the target located in the fourth-order reflection area, as shown in Figure 4 As shown, it can be seen that the simulation value and the theoretical value are in good agreement, thereby verifying the correctness of the multipath model constructed by the present invention.

[0072] Table 1 Lengths of ten two-way transmission paths

[0073]

[0074] Step 2: Identify potential target areas:

[0075] Use MTI to eliminate static clutter from the received echo, then use CFAR to extract the echo's ToA. This value is stored in vector T. The first value, τ1, is extracted as the two-way diffraction propagation delay.

[0076] ||C0P||=(τ1c-2||RC0||) / 2=r p (15)

[0077] △R u Radar measurement error △R u =0.5m, then the target potential area is:

[0078]

[0079] Step 3: Target potential area mesh division:

[0080] Divide the entire non-line-of-sight area into grids with a side length of 0.03m. If the center coordinate of the grid is located in the area Ω obtained in step 2, calculate its theoretical multipath according to the multi-level propagation model obtained in step 1 and store it in vector T s middle.

[0081] Step 4: Get the real target location

[0082] Traverse all grids and obtain their corresponding matching factors and error factors. The grid with the largest matching factor and the smallest error factor is selected and considered to be the true position of the target.

[0083] Taking the target of the 45th cycle and the target of the 96th cycle as an example, the results are as follows Figure 5 (a) and Figure 5 As shown in (b), it can be seen from the figure that the target is indeed located in the delineated target potential area. Specifically, the target positioning results are shown in Table 2.

[0084] Table 2 Simulation results of two cycles

[0085] Cycle 45 Cycle 96 Area Third-order reflection area Fifth-order reflection area Real location [2.89m,7.00m] [3.92m,11.64m] Extract Location [2.91m,7.02m] [3.96m,11.67m] error 0.291m 0.0529m

[0086] The positioning results of all 100 cycles are as follows Figure 5 As shown in (c), it can be seen that the extracted target position and the target real position are in good agreement. Specifically, the error between the real target position and the extracted target position at different periods is as follows: Figure 5 As shown in (d), the calculated average value is 0.0489 m. The running time of the present invention is 23.5 seconds. Therefore, the algorithm proposed in the present invention ensures accurate positioning while requiring less computation.

[0087] Those skilled in the art will appreciate that the embodiments described herein are intended to aid the reader in understanding the principles of the present invention, and it should be understood that the scope of the present invention is not limited to such specific descriptions and embodiments. Various modifications and variations are readily apparent to those skilled in the art. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are intended to be included within the scope of the claims.

Claims

1. A two-stage target positioning method for a long L-shaped building layout, characterized in that: The positioning method is based on an application scenario including: wall 1, wall 2, wall 3, a single-transmitter single-receiver radar, and a hidden target. Wall 1 and wall 2 form an L-shaped wall, and wall 2 and wall 3 are parallel. The single-transmitter single-receiver radar is set in front of wall 1, and the hidden target is located in the non-line-of-sight area between wall 2 and wall 3. The positioning method specifically includes the following steps: S1. Divide the non-line-of-sight area into reflection areas of different orders based on ray tracing, and then analyze the multipath propagation models in different reflection areas respectively; S2. Determine the target potential area based on the diffraction delay in the echo received by the single-transmitter, single-receiver radar according to the constructed multipath propagation model; S3, dividing the target potential area into grids according to the radar range resolution, and calculating the corresponding theoretical multipath delay; S4. Match the arrival time of the received echo with the theoretical multipath delay of the grid to obtain the true position of the target.

2. The two-stage target positioning method under a long L-shaped building layout according to claim 1 is characterized in that: When analyzing the multipath propagation model in different reflection areas, the one-way reflection path is divided into a simple one-way path and a compound two-way path.

3. The two-stage target positioning method under a long L-shaped building layout according to claim 2 is characterized in that: The multipath propagation models in different reflection areas are as follows: Consider four one-way propagation paths of electromagnetic waves, including diffraction path, first-order reflection path, second-order reflection path, and third-order reflection path; In the first-order reflection region, the diffraction path, the first-order reflection path, the second-order reflection path, and the third-order reflection path are all simple reflection paths; In the second-order reflection region, the diffraction path and the first-order reflection path are both composite two-way paths, and the second-order reflection path and the third-order reflection path are both simple reflection paths; In the third-order reflection region, the diffraction path, the first-order reflection path, and the second-order reflection path are all composite two-way paths, and the third-order reflection path is a simple reflection path; In the reflection region above the fourth order, the diffraction path, the first-order reflection path, the second-order reflection path and the third-order reflection path are all composite two-way paths.

4. The two-stage target positioning method under a long L-shaped building layout according to claim 3 is characterized in that: Step S2 specifically includes the following sub-steps: S21. Use moving target indication technology to eliminate the influence of static background echo and antenna coupling in the received echo signal; S22. Use a constant false alarm detector to extract the ToA of different paths in the echo and store them in vector T in ascending order: S23, based on the shortest two-way diffraction path distance, extract the minimum ToA value from vector T as the two-way diffraction propagation delay; S24. The distance between the corner and the target has the following relationship: ||C0P||=(τ1c-2||RC0||) / 2=r p Where C0 represents the horizontal coordinate of the corner point of the L-shaped wall, P represents the coordinate of the hidden target, τ1 represents the minimum ToA value in the vector T, c represents the propagation speed of electromagnetic waves in the air, R represents the coordinate of the single-transmitter and single-receiver radar, ||·|| represents the second norm, and r p The distance between the corner and the hidden target; S25, the area where the hidden target is located is denoted as Ω, and the expression is: Oh:{r min ≤r p ≤r max } Among them, △R u represents the radar measurement error, r max and r min They represent the upper and lower bounds of the potential area of ​​the hidden target, respectively.

5. The two-stage target positioning method under a long L-shaped building layout according to claim 4 is characterized in that: The theoretical multipath delay calculation formula corresponding to each grid is: in, represents the propagation delay of the mth transmission path in the k-order reflection region, R represents the propagation delay of the nth receiving path in the k-th order reflection region, m Indicates the radar virtual mirror position corresponding to the mth transmission path, R n The radar virtual mirror position corresponding to the nth receiving path, C m C represents the virtual mirror image position of the corner point of the L-shaped wall corresponding to the mth transmission path on the horizontal plane, n The virtual mirror image position of the corner point of the L-shaped wall corresponding to the nth receiving path on the horizontal plane.

6. The two-stage target positioning method under a long L-shaped building layout according to claim 5 is characterized in that: Step S4 is specifically as follows: The two-way theoretical transmission delay corresponding to each grid in step S3 is stored in the vector T s middle; Construct matching factor N s and error factor E s , N s Indicates T s The number of ToAs that successfully match in T, E s represents the sum of the corresponding matching errors; in, δ represents an empirical threshold with a smaller value, Represents the grid G s The round-trip transmission delay of Traverse all grids in the potential area of ​​the hidden target and store all the matching factors and error factors obtained in two vectors: Where, s∈{1,2,…,S}, S is the number of grids, and s is the index value; N s The largest grid corresponds to E s The smallest one is the grid where the real target is located.

Images