AoA assisted direct path tof super-resolution estimation method in multipath environment

Abstract

The application discloses an AOA-assisted direct wave TOF super-resolution estimation method in a multipath environment. First, radar data is collected, phase offsets are introduced between different channels and subcarriers through AOA and TOF, a wideband array signal model is established, then an estimation grid is divided within the AOA and TOF estimation range, a sparse constraint optimization model is constructed according to the sparsity of the real AOA-TOF compared with the estimation grid, electromagnetic waves penetrating through a building environment are effectively estimated through a variable resolution estimation method, variable resolution solving is completed, finally, based on the difference between the direct wave and the multipath in the AOA-TOF spectrum, the direct wave signal is extracted from all the paths estimated according to the direct wave AOA being equal to 0, and the direct wave TOF estimation is completed. The method can realize super-resolution estimation of the direct wave signal in the electromagnetic wave propagation path in a complex unknown scene, has the advantages of high resolution, small calculation amount, high direct wave separation accuracy and the like, and can be directly applied to CT building layout, wall-penetrating radar and other equipment.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of radar parameter super-resolution estimation, and particularly relates to an AOA auxiliary direct wave TOF super-resolution estimation method in a multipath environment. BACKGROUND

[0002] Direct wave Time of flight (TOF) super-resolution estimation method is widely studied because it can provide a fine-grained description of electromagnetic wave propagation to provide information support in indoor positioning, through-wall radar, building layout imaging and other fields. For the Computed Tomography (CT) building layout imaging technology, accurate direct wave TOF super-resolution estimation can reduce the demand of the system on the hardware bandwidth and improve the estimation accuracy of the key parameter direct wave TOF. Therefore, it has attracted great attention in recent years.

[0003] Many research institutions at home and abroad have carried out research on direct wave TOF estimation, and have produced rich research results. Direct wave TOF estimation is a signal processing technology, which estimates the TOF of electromagnetic waves by the different phases caused by the TOF of electromagnetic waves on different subcarriers. And according to the difference between direct wave and multipath, a direct wave separation method is designed. Direct wave TOF super-resolution estimation has related research work in many fields. The literature "Chen J, Zhang Y, Li H, et al. Ultrawideband Tomographic Imaging in Multipath-Rich Environment. IEEE Geoscience and Remote Sensing Letters, vol. 19, pp. 1-5, 2022" designs a cross-correlation-based direct wave separation method according to the propagation characteristics of multiple paths. The literature "Li N, Chen J, Guo S, et al. Building Layout Tomographic Imaging With Improved Delay Estimation Algorithm. IEEE Antennas and Wireless Propagation Letters, vol. 22, no. 5, pp. 1079-1083, May 2023" uses the formality of direct wave on adjacent scanning points to design a direct wave separation method. But the direct wave TOF estimation in the above method is severely dependent on the signal bandwidth, and the distinction between direct wave and multipath only depends on the time delay estimation result, which leads to high hardware demand and limited estimation quality of direct wave TOF estimation. The method of extracting direct wave by using the similarity of direct wave signals in adjacent sampling points is essentially also only using the difference between direct wave and multipath in TOF. The existing methods often rely on the experience of TOF distribution in distinguishing direct wave and multipath, such as the direct wave has larger energy compared with multipath, the TOF of direct wave between adjacent points is similar, etc. The information leads to the difficulty of improving the accuracy of these methods in distinguishing direct wave and multipath. And the existing method is severely dependent on the signal bandwidth, which leads to high hardware demand. Therefore, it is of great significance to study an Angle of Arrival (AOA) assisted direct wave TOF estimation method suitable for complex multipath environment. SUMMARY

[0004] To solve the above technical problems, the present application provides an AOA assisted direct wave TOF super-resolution estimation method in a multipath environment, which can effectively super-resolve electromagnetic wave propagation path parameters and improve the accuracy of direct wave path estimation.

[0005] The technical scheme adopted by the application is as follows:

[0006] S1, data acquisition is performed, a bistatic radar is arranged on the two sides of an unknown area in opposite directions, transmission detection is performed, an electromagnetic wave is emitted by a transmitting antenna to penetrate the unknown area, and a wideband array signal is collected by a receiving end array antenna as a receiving signal;

[0007] S2, based on the data collected in step S1, phase shifts are introduced between different channels and subcarriers for AOA and TOF, and then a wideband array signal model is established;

[0008] S3, an estimation grid is divided within the AOA and TOF estimation range, and a sparse constraint optimization model is constructed according to the sparsity of the real AOA-TOF compared with the estimation grid;

[0009] S4, the electromagnetic wave penetrating the building environment is effectively estimated by using a variable resolution estimation method, and variable resolution solving is completed;

[0010] S5, based on the difference between the direct wave and the multipath in the AOA-TOF spectrum, the direct wave signal is extracted from all the paths estimated according to the direct wave AOA being equal to 0, and the direct wave TOF estimation is completed.

[0011] Further, in step S2, the wideband array signal model is as follows:

[0012] It is assumed that the receiving end uses a linear uniform array (ULA), the distance between the antennas is d, the frequency of each frequency point in the wideband array signal is f k , the frequency difference between each frequency point is Δf, there are M antennas and K frequency points. It is assumed that the AOA of any one propagation path is θ, and the TOF is τ.

[0013] The phase difference introduced by AOA θ on the kth frequency point of the mth channel is The phase difference introduced by AOA θ on adjacent frequency points of all channels is The phase difference introduced by TOF τ on adjacent frequency points of all channels is φ τ = exp (-j2πΔfτ).

[0014] Wherein, Δf represents the phase difference between adjacent frequency points; and c represents the speed of light.

[0015] The phase difference between the kth frequency point on the mth antenna and the first frequency point on the first antenna is Considering all L echoes, the receiving signal on the kth frequency point on the mth antenna is represented as:

[0016] Wherein, al θ represents the amplitude of the echo along the l-th propagation path; l and τ l Let AOA and TOF represent the propagation path l, respectively. Then, the turning vector f(θ) corresponding to the l-th path is... l ,τ l ) is represented as: f(θ) l ,τ l )=[1,Φ 12 (θ l ,τ l ),…,Φ 1K (θ l ,τ l ),…,Φ MK (θ l ,τ l )] T .

[0017] in,[] T This represents the matrix transpose operation, where M and K represent the number of channels and frequency points of the broadband array signal, respectively.

[0018] Considering all paths, combine all steering vectors to form a steering matrix: F = [f(θ1,τ1),f(θ2,τ2),…, ...f(θ1,τ2),f(θ1,τ2),f(θ1 L ,τ L The received signals are stacked into an MK×1 vector y(t) = [y 11 ,y 12 ,…,y 1K ,…,y MK ] T .

[0019] Where y(t) represents the broadband array signal received at time t, y mk Let represent the frequency response at frequency k in the m-th channel. Then the received signal of a single frame is expressed as: y(t) = Fs(t) + n(t).

[0020] Among them, s(t)=[α1(t),α2(t),…,α L (t)] T Indicates the signal source, α l (t) represents the signal strength on the l-th path at time t, and n(t) = [n1(t), n2(t), ..., n MK (t)] T n represents the additive noise component in the received signal. mk (t) represents the additive noise at the k-th frequency point of the m-th channel.

[0021] At different times t1, t2, ..., t Z The received signal is represented as a matrix The wideband array signal model is represented as:

[0022] Y = FS + N

[0023] where Z represents the received data of Z frames, S = [s(t1), s(t2), …, s(t Z )] represents the signal source at all time points, and N represents noise.

[0024] Further, the step S3 is specifically as follows:

[0025] The estimation grid grid = [(θ1, τ1), …, (θ L , τ L )] is divided within the AOA and TOF estimation range, i.e., [θ1, θ L ] and [τ1, τ L ], and a steering matrix F is constructed. The sparse constraint optimization model is represented as:

[0026] where represents the signal source matrix obtained by estimation; represents the square of the Frobenius norm; and β represents the reconstruction error allowed in the model.

[0027] Further, the step S4 is specifically as follows:

[0028] First, the estimation grid is divided with a large step size L1 represents the number of units in the coarse-grained division grid. The steering matrix F1 corresponding to the estimation grid grid1 is constructed, and a coarse-grained model is constructed. The coarse-grained estimation is obtained by solving the model.

[0029] Then, the estimation grid is divided with a small step size near the target parameter region L2 represents the number of units in the fine-grained division grid, and a fine-grained model is constructed. The fine-grained estimation is obtained by solving the model.

[0030] The beneficial results of this invention are as follows: The method first acquires radar data, introduces phase shifts between different channels and subcarriers using AOA and TOF, and establishes a broadband array signal model. Then, it divides the estimation grid within the AOA and TOF estimation range, constructs a sparse constraint optimization model based on the sparsity of the actual AOA-TOF compared to the estimated grid, and effectively estimates electromagnetic waves penetrating the building environment using a variable resolution estimation method, completing the variable resolution solution. Finally, based on the difference between direct waves and multipath signals in the AOA-TOF spectrum, it extracts the direct wave signal from all estimated paths based on the direct wave AOA being equal to 0, simultaneously completing the direct wave TOF estimation. This method can achieve super-resolution estimation of direct wave signals in electromagnetic wave propagation paths in complex and unknown scenarios, effectively suppressing multipath interference and solving problems such as low resolution in electromagnetic wave parameter estimation and difficulty in separating direct waves from multipath signals. It has advantages such as high resolution, low computational load, and high accuracy in direct wave separation, and can be directly applied to equipment such as CT-like building layouts and through-wall radar. Attached Figure Description

[0031] Figure 1 This is a flowchart of an AOA-assisted direct-wave TOF super-resolution estimation method under multipath environment according to the present invention.

[0032] Figure 2 This is a schematic diagram of radar node operation in an embodiment of the present invention.

[0033] Figure 3 This is a diagram showing the coarse-grained estimation results in an embodiment of the present invention.

[0034] Figure 4 This is a diagram showing the fine-grained estimation results in an embodiment of the present invention.

[0035] Figure 5 This is a time-domain diagram of the received signal strength of the radar scanning node at a 0-degree angle in an embodiment of the present invention.

[0036] Figure 6 This is a diagram showing the direct wave delay estimation results for all radar scanning nodes received signals in this embodiment of the invention. Detailed Implementation

[0037] The method of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0038] like Figure 1 The flowchart of the AOA-assisted direct wave TOF super-resolution estimation method in a multipath environment according to the present invention is shown below. The specific steps are as follows:

[0039] S1, data acquisition is performed, a transmitting and receiving dual-base radar is arranged on two sides of an unknown area in opposite directions, transmission detection is performed, an electromagnetic wave is emitted by a transmitting antenna to penetrate the unknown area, and a wideband array signal is collected by a receiving end array antenna after the electromagnetic wave to serve as a receiving signal;

[0040] S2, based on the data collected in step S1, AOA and TOF introduce phase offsets between different channels and subcarriers, and then a wideband array signal model is established;

[0041] S3, an estimation grid is divided in an AOA and TOF estimation range, and a sparse constraint optimization model is constructed according to the sparsity of a real AOA-TOF compared with the estimation grid;

[0042] S4, an electromagnetic wave penetrating a building environment is effectively estimated by a variational resolution estimation method, and variational resolution solving is completed;

[0043] S5, based on the difference between a direct wave and multipath in an AOA-TOF spectrum, a direct wave signal is extracted from all paths estimated according to the direct wave AOA being equal to 0, and direct wave TOF estimation is completed.

[0044] In the embodiment, the step S1 is specifically as follows:

[0045] A radar node working schematic diagram in the embodiment is shown in Figure 2 The radar node includes an omnidirectional transmitting antenna and a group of receiving end array antennas, and the transmitting and receiving nodes are arranged on two sides of an unknown area in opposite directions, and the receiving end array antennas receive transmission signals after scene interaction.

[0046] The wideband array signal of the receiving end is denoted as Y, which is composed of multiple frames of collected data, that is, Y = [y(t1), y(t2), …, y(tN)]. Z )].

[0047] Each frame of the receiving signal is composed of measurement values y mk (t) on each channel and a frequency point in the frame, that is, y(t) = [y 11 (t), y 12 (t), …, y 1K (t), …, y MK (t)] T , denotes a column vector, and the measurement value is denoted as

[0048] In the embodiment, in the step S2, the wideband array signal model is specifically as follows:

[0049] The receiving end is set to adopt a uniform linear array (ULA), the interval between each antenna is d, the frequency of each frequency point in the wideband array signal is f k , the frequency difference between each frequency point is Δf, there are M antennas and K frequency points. It is assumed that the AOA of any one propagation path is θ and the TOF is τ.

[0050] The AOA θ introduces a phase difference on the mth channel and the kth frequency point The AOA θ introduces a phase difference on all channels and adjacent frequency points The TOF τ introduces a phase difference φ on all channels and adjacent frequency points τ = exp (-j2πΔfτ).

[0051] Wherein, Δf represents the phase difference between adjacent frequency points; c represents the speed of light.

[0052] The phase difference between the kth frequency point on the mth antenna and the first frequency point on the first antenna is Considering all L echoes, the received signal on the kth frequency point on the mth antenna is represented as:

[0053] Wherein, a l represents the amplitude of the echo on the lth propagation path; θ l and τ l represent the AOA and TOF of the lth propagation path, respectively. The steering vector f(θ l ,τ l ) corresponding to the lth path is represented as: f(θ l ,τ l ) = [1, Φ 12 (θ l ,τ l ), …, Φ 1K (θ l ,τ l ), …, Φ MK (θ l ,τ l )] T .

[0054] Wherein, [] T represents the transpose operation of the matrix, and M and K represent the number of channels and the number of frequency points of the wideband array signal, respectively.

[0055] Considering all paths, all steering vectors are combined to form a steering matrix: F = [f(θ1, τ1), f(θ2, τ2), …, f(θ L ,τ L )]. The received signal is stacked into an MK × 1 vector y(t) = [y11 y 12 ,…,y 1K ,…,y MK ] T .

[0056] where y(t) represents the received wideband array signal at time t, y mk represents the frequency response at the mth channel and kth frequency bin. The received signal of a single frame is represented as: y(t) = Fs(t) + n(t).

[0057] where s(t) = [α1(t), α2(t), …, α L (t)] T represents the signal source, α l (t) represents the signal strength of the lth path at time t, n(t) = [n1(t), n2(t), …, n MK (t)] T represents the additive noise component in the received signal, n mk (t) represents the additive noise at the mth channel and kth frequency bin.

[0058] The received signals at different times t1, t2, …, t Z are represented as a matrix The wideband array signal model is represented as:

[0059] Y = FS + N

[0060] where Z represents the received data of Z frames, S = [s(t1), s(t2), …, s(t Z )] represents the signal source at all time points, and N represents the noise.

[0061] In this embodiment, the step S3 is specifically as follows:

[0062] According to the fact that the number of real echo paths is limited and the real AOA and TOF are sparse compared to the AOA and TOF estimation grid, a sparse constraint is used to construct an optimization model to perform super-resolution estimation of the AOA and TOF.

[0063] Within the estimation ranges of the AOA and TOF, i.e., [θ1, θ L ] and [τ1, τ L ], an estimation grid grid = [(θ1, τ1), …, (θ L , τ L )] is divided and a steering matrix The sparse constraint optimization model is represented as:

[0064] where, represents a signal source matrix obtained by estimation; represents the square of the Frobenius norm; β represents the reconstruction error allowed in the model.

[0065] In the embodiment, the step S4 is specifically as follows:

[0066] The estimation method of variable resolution reduces the operation amount on the basis of ensuring fine-grained estimation of the region of interest.

[0067] First, the estimation grid is divided in large steps L1 represents the number of units in the grid of coarse-grained division. The steering matrix F1 is constructed according to the grid and constitutes the coarse-grained model represents the steering matrix corresponding to the estimation grid grid1. The solution of the model can obtain the coarse-grained estimation Then, the estimation grid is divided in small steps around the target parameter region on this basis L2 represents the number of units in the grid of fine-grained division, on the basis of which the fine-grained model is constructed represents the steering matrix corresponding to the estimation grid grid2, and the solution of the model can obtain the fine-grained estimation

[0068] In the embodiment, in the step S5, the difference between the direct wave and the multipath in AOA is specifically as follows:

[0069] The bistatic radar is oppositely distributed on both sides of the unknown region, the multipath is caused by single or multiple reflections of the wall in the unknown region, and the direct wave is generated by transmitting the wall. Therefore, the direct wave and the multipath have obvious and stable difference in AOA.

[0070] The bistatic radar is oppositely arranged, the AOA of the direct wave is 0, and the AOA of the multipath formed due to wall reflection, refraction, etc. is not 0.

[0071] In the embodiment, the conductivity, relative permittivity and size of different materials in the simulation scene are shown in Table 1:

[0072] Table 1

[0073] Material Relative dielectric constant Electrical conductivity Size Wood 3 0.1 2.06 m * 2.06 m, wall width 0.15 m Bricks 6 0.1 0.3 m * 1.76 m

[0074] The arrangement of the radar nodes is shown in Figure 2 The coarse-grained estimation result is shown in Figure 3 It can be found that the direct wave TOF is in the range of 16.667-20 nanoseconds, and further fine-grained estimation of TOF in the range can obtain the direct wave TOF of 18 nanoseconds. If direct fine-grained estimation is performed, the result is shown in Figure 4It can be seen that the direct wave TOF is 18 nanoseconds.

[0075] When the radar node moves on the two scanning tracks of 0 degrees and 90 degrees, multiple direct wave estimation results can be obtained. Taking the 0-degree scanning track as an example, the time-domain echoes on all measurement points are as shown in the figure. Figure 5 It can be seen that multipath exists on all measurement points, and the direct wave does not necessarily satisfy the empirical information that it has greater energy or smaller TOF compared with multipath. Moreover, the TOF of the direct wave between adjacent scanning points is close, which is not true for all scanning points. The existing MAE algorithm and Improved-MAE method are used to extract the direct wave TOF, and the estimation results of the method in the present application on the two scanning tracks of 0 degrees and 90 degrees are compared as shown in the figure. Figure 6 It can be seen that the method in the present application can more stably separate the direct wave and has higher TOF estimation accuracy. The simulation results above show that the method in the present application can realize low-computational super-resolution estimation of electromagnetic wave parameters, stable separation of direct waves, and has the characteristics of low hardware requirement and strong environmental adaptability.

[0076] In summary, to solve the problems of difficulty in extraction of direct waves and high resolution requirement of electromagnetic wave multipath propagation in the prior art, the method in the present application can effectively jointly estimate the AOA and TOF parameters of electromagnetic waves, establish an AOA-assisted direct wave extraction method based on the difference between direct waves and multipath in the AOA domain, and then extract direct waves in a complex multipath environment. The method in the present application can effectively extract direct waves in a complex multipath environment under the condition of limited bandwidth, while ensuring limited computational complexity, realize super-resolution estimation of direct wave signals in the electromagnetic wave propagation path in a complex unknown scene, effectively suppress the interference of multipath, and solve the problems of low resolution of electromagnetic wave parameter estimation and difficulty in separation of direct waves from multipath.

[0077] Those skilled in the art will appreciate that the embodiments described herein are presented for the purpose of helping the reader understand the principles of the present application and should be understood as not limiting the scope of protection of the present application to such specific statements and embodiments. The present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the scope of protection of the claims of the present application.

Claims

1. An AOA-assisted direct wave TOF super-resolution estimation method in a multipath environment, the specific steps being as follows: S1, data acquisition, a bistatic radar is arranged on both sides of an unknown area in opposite directions, transmission detection is performed, an electromagnetic wave is emitted by a transmitting antenna to penetrate the unknown area, and a wideband array signal is collected by a receiving end array antenna as a receiving signal; S2, based on the data collected in step S1, phase offsets are introduced between different channels and subcarriers for AOA and TOF, and then a wideband array signal model is established; S3, an estimation grid is divided within the AOA and TOF estimation range, and a sparse constraint optimization model is constructed according to the sparsity of the real AOA-TOF compared with the estimation grid; S4, the electromagnetic wave penetrating the building environment is effectively estimated by a variable resolution estimation method, and variable resolution solving is completed; S5, based on the difference between the direct wave and the multipath in the AOA-TOF spectrum, the direct wave signal is extracted from all the paths estimated according to the direct wave AOA being equal to 0, and the direct wave TOF estimation is completed.

2. The AOA-assisted direct-path TOF super-resolution estimation method in a multipath environment according to claim 1, characterized in that, In the step S2, the wideband array signal model is specifically as follows: The receiving end adopts a linear uniform array (ULA), the interval between each antenna is , the frequency of each frequency point in the wideband array signal is , the frequency difference between each frequency point is , there are antennas, frequency points; the AOA of any one propagation path is , and the TOF is ; By AOA In the first On the first Introducing phase difference By AOA Introducing phase difference on adjacent frequency points of all channels ; By TOF Introducing phase difference on adjacent frequency points of all channels ; wherein denotes the speed of light; No. The first antenna The phase difference between each frequency point and the first frequency point of the first antenna is Consider all The first echo, then the second echo... The first on the root antenna The received signal at each frequency point is represented as: ; wherein, denotes the amplitude of the echo on the mth propagation path; and denotes the AOA and TOF, respectively, of the mth propagation path; then the steering vector corresponding to the mth path is denoted by and is given by: ; wherein, denotes a transpose operation of a matrix, and denotes the number of channels and the number of frequency bins of the wideband array signal, respectively. Considering all paths, all steering vectors are combined to form a steering matrix: ; stack the received signals into a vector ; wherein represents a wideband array signal received at a time instant, represents the frequency response at the mth channel of the frequency bin; the received signal of a single frame is represented as: ; wherein represents a signal source, represents the time instant, the signal strength on the path, represents an additive noise component in the received signal, represents the additive noise on the channel on the frequency bin; The received signals at different times are represented as matrices The wideband array signal model is then represented as: ； wherein represents a co frame receiving data, represents a signal source at all time points, represents noise.

3. The AOA-assisted direct-path TOF super-resolution estimation method in a multipath environment according to claim 2, characterized in that, The step S3 is specifically as follows: In the AOA and TOF estimation range, i.e. and dividing the estimation grid and forming a steering matrix The sparse constraint optimization model is represented as: ; wherein , denotes the signal source matrix obtained by estimation; denotes the square of the Frobenius norm; denotes the reconstruction error allowed in the model.

4. The AOA-assisted direct-path TOF super-resolution estimation method in multipath environment according to claim 2, characterized in that, The step S4 is specifically as follows: First, the estimation grid is divided with a large step size , denotes the number of cells in the coarse grid; a steering matrix is constructed from this grid and forms the coarse model , denotes the steering matrix corresponding to the estimation grid , the solution of which yields the coarse estimate ; Then, the estimation grid is divided in small steps around the target parameter region. , This represents the number of elements in the fine-grained mesh, which is then used to construct the fine-grained model. , Representation and estimation grid The corresponding turning matrix is ​​used to solve the model and obtain a fine-grained estimate. .

Images