Single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition

Abstract

The application relates to the technical field of signal processing, in particular to a single-bit single-shot signal direction of arrival estimation method based on Hankel matrix decomposition, which comprises the following steps: acquiring a received single-shot sampling signal and a corresponding single-bit quantization signal; constructing an initial DOA estimation model based on the sign consistency constraint between the single-shot sampling signal and the single-bit quantization signal and the Hankel low-rank constraint of the single-shot sampling signal; optimizing the initial DOA estimation model by using a Hankel matrix decomposition and function smoothing method to obtain an optimized DOA estimation model; solving the optimized DOA estimation model by using a momentum gradient descent algorithm to obtain a reconstructed original signal; and processing the reconstructed original signal by using a single-shot subspace method to finally obtain the estimated value of the direction of arrival of each signal source. The method improves the precision, efficiency and robustness of the direction of arrival estimation.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of signal processing, in particular to a single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition. BACKGROUND

[0002] Array signal processing is an important branch of modern signal processing, and has a wide range of applications in radar detection, wireless communication, underwater sonar, etc. In a number of programmatic documents, it is clearly stated that the key core technology bottleneck of array signal processing should be broken through as soon as possible to promote the innovative development of information sensing technology.

[0003] Among the many tasks of array signal processing, the direction of arrival (DOA) estimation is one of the core problems, and its goal is to estimate the incident direction of the target according to the signals received by the sensor array.

[0004] Most of the current DOA estimation methods rely on multi-snapshot sampling and high-bit analog-to-digital converters. Multi-snapshot sampling requires multiple-time, multi-frame signal covariance matrix estimation, which has the problem of insufficient real-time performance in transient target detection and high-speed motion scenarios. High-bit analog-to-digital converters have the problems of hardware complexity and high power consumption, which are not suitable for low-power or distributed sensing systems.

[0005] Single-bit sampling technology only retains the sign information (positive or negative) of the signal, greatly reducing the hardware complexity and data storage requirements. However, the current single-bit single-snapshot DOA estimation method has the following shortcomings.

[0006] First, it relies on grid-based spectrum compressed sensing, which has grid mismatch error and low estimation accuracy.

[0007] Second, the optimization model is often non-convex and non-smooth, which leads to slow convergence speed of related algorithms and easy to fall into the dilemma of local optimum.

[0008] Therefore, there is an urgent need in the art for a DOA estimation method that can maintain high accuracy and reduce computational complexity under single-bit single-snapshot sampling conditions. SUMMARY

[0009] To solve the above technical problems, embodiments of the present application propose a single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition, which uses single-bit sampling to effectively reduce hardware complexity and power consumption, perfectly adapts to single-snapshot observation conditions, improves the accuracy and robustness of direction of arrival estimation, avoids falling into the dilemma of local optimum, speeds up the convergence speed, and improves the efficiency of direction of arrival estimation.

[0010] To achieve the above object, embodiments of the present application provide a single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition, which comprises the following steps: obtaining a single-snapshot sampling signal received by a vehicle-mounted millimeter wave radar array and a single-bit quantized signal corresponding to the single-snapshot sampling signal; constructing an initial DOA estimation model based on a sign consistency constraint between the single-snapshot sampling signal and the single-bit quantized signal and a Hankel low-rank constraint of the single-snapshot sampling signal; optimizing the initial DOA estimation model by using a Hankel matrix decomposition and a function smoothing method to obtain an optimized DOA estimation model; solving the optimized DOA estimation model by using a momentum gradient descent algorithm to realize reconstruction of the original signal and obtain a reconstructed original signal; and finally obtaining an estimated value of the direction of arrival of each signal source by processing the reconstructed original signal by using a single-snapshot subspace method.

[0011] To achieve the above object, embodiments of the present application also provide a single-bit single-snapshot signal direction of arrival estimation system based on Hankel matrix decomposition, which comprises: an acquisition module configured to acquire a single-snapshot sampling signal received by a vehicle-mounted millimeter wave radar array and a single-bit quantized signal corresponding to the single-snapshot sampling signal; a construction module configured to construct an initial DOA estimation model based on a sign consistency constraint between the single-snapshot sampling signal and the single-bit quantized signal and a Hankel low-rank constraint of the single-snapshot sampling signal; an optimization module configured to optimize the initial DOA estimation model by using a Hankel matrix decomposition and a function smoothing method to obtain an optimized DOA estimation model; an original signal reconstruction module configured to solve the optimized DOA estimation model by using a momentum gradient descent algorithm to realize reconstruction of the original signal and obtain a reconstructed original signal; and an estimation execution module configured to finally obtain an estimated value of the direction of arrival of each signal source by processing the reconstructed original signal by using a single-snapshot subspace method.

[0012] To achieve the above object, embodiments of the present application also provide an electronic device, which comprises a processor and a memory, wherein the memory stores instructions executable by the processor, and the processor is configured to execute the instructions so that the electronic device can implement a single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition as described above.

[0013] To achieve the above object, embodiments of the present application also provide a computer-readable storage medium storing a computer program, wherein the computer program is executable by a processor to implement a single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition as described above.

[0014] The embodiment of the application proposes a single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition, which brings the following beneficial effects.

[0015] First, the hardware complexity and power consumption are reduced. The application adopts a single-bit sampling method, only the signal symbol information is retained to complete the processing, avoiding the dependence on high-precision analog-to-digital converters (ADCs), thereby significantly reducing the hardware cost and system energy consumption.

[0016] Second, it adapts to single-snapshot observation conditions. The application fully excavates the structural characteristics of single-snapshot signals by constructing a Hankel matrix and introducing a low-rank constraint, effectively solving the problem that traditional methods cannot achieve stable direction of arrival estimation under single-snapshot conditions.

[0017] Third, the precision and robustness of direction of arrival estimation are improved. The application introduces a sign consistency constraint in model construction and combines Hankel matrix decomposition and function smoothing methods for optimization processing, which can ensure the consistency between the reconstructed signal and the observed signal, thereby maintaining high estimation accuracy and robustness in a low signal-to-noise ratio environment.

[0018] Fourth, the computing efficiency is improved, and real-time processing is facilitated. The application uses a momentum gradient descent algorithm to solve the optimization model, which can accelerate the convergence speed and avoid falling into local optimum, and has lower computational complexity than traditional convex optimization methods, making it more suitable for real-time signal processing scenarios.

[0019] Optionally, the vehicle-mounted millimeter wave radar array is a uniform linear array, which is composed of a total of array elements, is an integer greater than 1, and the array element spacing is The single-snapshot sampling signal received by the vehicle-mounted millimeter wave radar array is which is expressed by the formula as:

[0020] ;

[0021] ;

[0022] ;

[0023] wherein represents a noise vector, , represents a vector space with a dimension of , and the upper right corner represents a transpose operation, represents the signal received by the i-th array element of the vehicle-mounted millimeter wave radar array, ​denotes the sparsity of the spectrum, i.e., the number of signal sources, denotes the amplitude of the echo of the th signal source, denotes the wavelength of the echo, denotes the direction of arrival of the th signal source, i.e., the direction of incidence of the th signal source;

[0024] denotes the single-snapshot spectrum-sparse signal , = , then Based on denotes:

[0025] ;

[0026] Based on , the single-bit quantized signal corresponding to , , is obtained by the formula:

[0027] ;

[0028] wherein, denotes taking the real part, denotes taking the imaginary part, denotes taking the sign element by element.

[0029] Optionally, the sign consistency constraint between the single-snapshot sampled signal and the single-bit quantized signal is represented by the formula:

[0030] ;

[0031] wherein, denotes the conjugate of , denotes the dot product operation;

[0032] Let denote the up-sampling operator of the Hankel matrix, is used to up-sample to , denotes the dimension of the vector space after up-sampling, then the Hankel low-rank constraint of the single-snapshot sampled signal is represented by the formula:

[0033] ;

[0034] wherein, denotes the up-sampling Hankel matrix of , denotes the i-th column of the Hankel matrix, denotes the i-th column of the Hankel matrix,

[0035] Based on the sign consistency constraint between the single-shot sampling signal and the single-bit quantized signal and the Hankel low-rank constraint of the single-shot sampling signal, an initial DOA estimation model is constructed, comprising:

[0036] The sign consistency constraint between the single-shot sampling signal and the single-bit quantized signal is used to make the sign of the to-be-reconstructed signal consistent with the sign of , and the Hankel low-rank constraint of the single-shot sampling signal is used to make the rank of consistent with the rank of ;

[0037] Based on , an initial DOA estimation model is constructed, and the initial DOA estimation model is represented as:

[0038] ;

[0039] wherein, denotes the single-sided norm, the upper right corner mark denotes taking power, denotes retaining the element with a negative absolute value and setting the element with a positive absolute value to zero, is a positive integer, when facing sparse noise including impulse noise, taking , when facing dense noise including white noise, taking , denotes the norm, and the upper right corner mark 2 denotes taking square.

[0040] Optionally, the initial DOA estimation model is optimized by using a Hankel matrix decomposition and a function smoothing method to obtain an optimized DOA estimation model, comprising:

[0041] The Hankel matrix decomposition is used to decompose , and is decomposed into , the upper right corner mark denotes taking a conjugate matrix operation, and are both iterative variables, , ;

[0042] In order to ensure ​​with Hankel low-rank structure, adding operator constraints, the operator constraints are represented as:

[0043] ;

[0044] ;

[0045] wherein, denotes a unit operator, denotes a linear operator, denotes an inverse operator of, denotes an adjoint operator of;

[0046] Based on the operator constraints, the initial DOA estimation model is converted into an intermediate DOA estimation model, and the intermediate DOA estimation model is represented as:

[0047] ;

[0048] The function is smoothed using a function smoothing method to smooth into , for the variable , satisfies:

[0049] ;

[0050] wherein, is a smoothing parameter, ;

[0051] On the basis of the intermediate DOA estimation model, the Hankel low-rank constraint of the single-shot sampling signal is converted into a penalty term, and an optimized DOA estimation model is obtained, and the optimized DOA estimation model is represented as:

[0052] ;

[0053] wherein, is a penalty parameter, , denotes an identity relationship.

[0054] Optionally, the momentum gradient descent algorithm is used to solve the optimized DOA estimation model to realize the reconstruction of the original signal, and the reconstructed original signal comprises:

[0055] The spectrum initialization processing of and is normalized to obtain initialized and ;

[0056] A momentum gradient descent algorithm is designed by combining momentum acceleration methods, and the momentum gradient descent algorithm is based on... and The optimized DOA estimation model was solved, and a total of [number] steps were performed. Round iteration, It is an integer greater than 1;

[0057] based on The output results after each iteration are used to reconstruct the original signal, resulting in the reconstructed original signal.

[0058] Optionally, for and Perform normalized spectral initialization to obtain the initialized spectrum. and ,include:

[0059] Using the following formula, Upgraded Hankel matrix Perform truncated singular value decomposition:

[0060] ;

[0061] in, express The rank approximation, , , ;

[0062] Based on the following formula, , , right and Perform normalized spectral initialization to obtain the initialized spectrum. and :

[0063] ;

[0064] ;

[0065] in, This represents the Frobenius norm normalization operator;

[0066] A momentum gradient descent algorithm is designed by combining momentum acceleration methods, and the momentum gradient descent algorithm is based on... and The optimized DOA estimation model is solved using the following formula:

[0067] ;

[0068] ;

[0069] ;

[0070] ;

[0071] wherein, represents a momentum parameter, represents a step size, represents the i-th iteration, represents the i-th iteration, and represent the gradient taken in the direction of and the gradient taken in the direction of, respectively; Based on the output results after i iterations,

[0072] and , the reconstruction of the original signal is performed to obtain the reconstructed original signal , which is expressed by the formula:

[0073] .

[0074] Optionally, set ;

[0075] When , and are updated by the following formula:

[0076] ;

[0077] ;

[0078] When , and are updated by the following formula:

[0079] ;

[0080] . BRIEF DESCRIPTION OF DRAWINGS

[0081] ​​​​​​​In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the description of the embodiments of the present application or the related art will be briefly introduced. Obviously, the following drawings are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor based on these drawings, and the drawings described herein are only used to explain the present application, and not to limit the present application.

[0082] Figure 1 is a flow chart of a single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition provided by an embodiment of the present application;

[0083] Figure 2 is a schematic diagram of a vehicle-mounted millimeter wave radar array receiving signal provided by an embodiment of the present application;

[0084] Figure 3 is a schematic diagram of a comparison result of different smoothing parameters provided by an embodiment of the present application,

[0085] Figure 4 is a schematic diagram of the relationship between the success rate and the signal-to-noise ratio of different algorithms under the conditions of and provided by an embodiment of the present application;

[0086] Figure 5 is a schematic diagram of the relationship between the root mean square error (RMSE) and the signal-to-noise ratio of different algorithms under the conditions of and provided by an embodiment of the present application;

[0087] Figure 6 is a structural schematic diagram of a single-bit single-snapshot signal direction of arrival estimation system based on Hankel matrix decomposition provided by another embodiment of the present application;

[0088] Figure 7 is a structural schematic diagram of an electronic device provided in another embodiment of the present application. DETAILED DESCRIPTION

[0089] ​​In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the embodiments of the present application will be described in detail below with reference to the drawings. Those skilled in the art can understand that, in the embodiments of the present application, many technical details are proposed in order to make the readers better understand. However, the technical solutions claimed by the present application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the following embodiments is for the convenience of description, and should not constitute any limitation on the specific implementation modes of the present application. The following embodiments can be combined and referenced with each other without contradiction.

[0090] One embodiment of the present application proposes a single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition, which is applied to a server. The implementation details of the single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition proposed in the present embodiment will be described in detail below. The following content only provides implementation details for the convenience of understanding, and is not necessary for implementing the present solution.

[0091] The specific process of the single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition proposed in the present embodiment can be as shown in Figure 1 , which includes:

[0092] Step 11, obtaining the single-snapshot sampling signal received by the vehicle-mounted millimeter wave radar array and the single-bit quantized signal corresponding to the single-snapshot sampling signal.

[0093] In a specific implementation, the direction of arrival estimation method proposed in the present embodiment is designed for a vehicle-mounted millimeter wave radar array. Therefore, when performing direction of arrival estimation, the single-snapshot sampling signal received by the vehicle-mounted millimeter wave radar array and the single-bit quantized signal corresponding to the single-snapshot sampling signal need to be obtained first.

[0094] In one example, as shown in Figure 2 , in the vehicle-mounted millimeter wave radar system of the intelligent driving vehicle, the vehicle needs to detect and locate the surrounding targets (such as pedestrians, vehicles or obstacles, etc.) in real time in a complex road environment. By estimating the incident direction of the echo signal of the target, the specific position of the target in space can be further calculated, thereby providing key support for environment perception and path decision.

[0095] The vehicle-mounted millimeter wave radar array is a uniform linear array (ULA), which is composed of array elements in total, is an integer greater than 1, the array element spacing is , and the single-snapshot sampling signal received by the vehicle-mounted millimeter wave radar array is which can be expressed by the formula:

[0096] ;

[0097] ;

[0098] ;

[0099] wherein, represents a noise vector, , represents a vector space with dimension , the upper right corner denotes the transpose operation, represents the signal received by the th array element of the vehicle-mounted millimeter wave radar array, represents the spectral sparsity, i.e., the number of signal sources, represents the amplitude of the echo of the th signal source, represents the wavelength of the echo, represents the direction of arrival of the th signal source, i.e., the incident direction of the th signal source, .

[0100] In order to reduce the requirements of vehicle-mounted devices in power consumption, cost and volume, the embodiment selects single-bit sampling through a comparator, and the single-shot spectral sparse signal is denoted as , = , then based on denoted as , based on , i.e., the single-bit quantized signal corresponding to , , is expressed by the formula:

[0101] ;

[0102] wherein, denotes taking the real part, denotes taking the imaginary part, denotes taking the sign element by element.

[0103] The specific meaning of the variable , can be represented as:

[0104] .

[0105] based on It can be determined that the target of the direction-of-arrival estimation in this embodiment is to utilize the ULA received... Estimate the direction of arrival of each signal source .

[0106] Step 12: Based on the symbol consistency constraint between the single-snapshot sampled signal and the single-bit quantized signal and the Hankel low-rank constraint of the single-snapshot sampled signal, construct the initial DOA estimation model.

[0107] In the specific implementation, after obtaining the single-snapshot sampling signal and its corresponding single-bit quantized signal, an initial DOA estimation model can be constructed based on the symbol consistency constraint between the single-snapshot sampling signal and the single-bit quantized signal and the Hankel low-rank constraint of the single-snapshot sampling signal.

[0108] First, we introduce the sign consistency between single-shot sampled signals and single-bit quantized signals, also known as the sign consistency constraint. When the signal-to-noise ratio is relatively high, noise... right The symbol has less influence, which makes it and The symbols are basically the same. Therefore, using and From the sign consistency, we can obtain the sign consistency constraint as follows:

[0109] ;

[0110] in, express conjugate, This indicates the dot product operation.

[0111] Next, we will introduce the Hankel low-rank constraint for single-shot sampled signals, i.e., the Hankel low-rank constraint. Let... Represents the dimension-raising operator of the Hankel matrix. Used to Upgrade to , This represents the dimension of the vector space after the dimension increase. The Hankel low-rank constraint of a single snapshot sampled signal is expressed by the formula:

[0112] ;

[0113] in, express The upgraded Hankel matrix, Represents the first of the Hankel matrix OK, Represents the first of the Hankel matrix List.

[0114] The sign consistency constraint is used to make the sign of the signal to be reconstructed as consistent as possible with the sign of , i.e., to minimize the number of elements with opposite signs. The Hankel low-rank constraint is used to make the rank of consistent with the rank of , i.e., to ensure that the rank of is also , .

[0115] Based on the above two constraints, the initial DOA estimation model can be constructed as:

[0116] ;

[0117] where denotes the one-sided norm, and the upper right superscript denotes taking the power, denotes retaining the elements with negative absolute values and setting the elements with positive absolute values to zero, is a positive integer, and when facing sparse noise including impulse noise, is taken as , and when facing dense noise including white noise, is taken as , denotes the norm, and the upper right superscript 2 denotes taking the square, The introduction of aims to limit the size of , and thus effectively control the range of the reconstructed signal.

[0118] It can be understood that the initial DOA estimation model does not strictly require the signs to be completely consistent, and thus has good fault tolerance for the sign flipping problem caused by the noise environment.

[0119] Step 13: The initial DOA estimation model is optimized by using the Hankel matrix decomposition and function smoothing method to obtain an optimized DOA estimation model.

[0120] In specific implementation, after the construction of the initial DOA estimation model is completed, the initial DOA estimation model needs to be optimized by using the Hankel matrix decomposition and function smoothing method, so as to obtain an optimized DOA estimation model.

[0121] The initial DOA estimation model is difficult to solve in polynomial time, so the Hankel low-rank constraint needs to be replaced by an equivalent form. To ensure the low-rank structure, the present embodiment selects to decompose by using Hankel matrix decomposition, and decomposition into , the upper right superscript of denotes the conjugate matrix operation, and are iteration variables, , . Through Hankel matrix decomposition, the parameter quantity is successfully reduced from to , effectively reducing the scale of updating parameters, and further improving the calculation efficiency.

[0122] In order to ensure has a Hankel low-rank structure, the operator constraint shown below is added in this embodiment:

[0123] ;

[0124] ;

[0125] wherein, denotes the identity operator, denotes the linear operator, denotes the inverse operator of , and denotes the adjoint operator of .

[0126] For the variable , satisfies , denotes the number of the th diagonal element in the dimensional matrix, .

[0127] Based on the operator constraint, the initial DOA estimation model can be converted into an intermediate DOA estimation model, which is represented as:

[0128] .

[0129] However, in the intermediate DOA estimation model, the operator is non-smooth, so solving the intermediate DOA estimation model is a non-smooth and non-convex optimization problem. The lack of smoothness makes it impossible to combine the momentum acceleration technique to design a solution algorithm for the intermediate DOA estimation model, which seriously affects the convergence performance of the algorithm.

[0130] In order to solve this problem, the function smoothing method is used to smooth into . For the variable , satisfies, , The rectified linear unit (ReLU) commonly used in deep learning is shown, and the smoothing operator of ReLU is used to The smoothing design of the operator uses the negative function of the SquarePlus function to perform smoothing.

[0131] For the variable , satisfies:

[0132] ;

[0133] wherein, is a smoothing parameter, when , degenerates to As shown in Figure 3 , approaches , The smaller the value of , The smaller the gap.

[0134] On the basis of the intermediate DOA estimation model, the Hankel low-rank constraint is converted into a penalty term, and the optimized DOA estimation model is obtained, and the optimized DOA estimation model is represented as:

[0135]

[0136] wherein, is a penalty parameter, , represents an identity relationship.

[0137] Step 14, using the momentum gradient descent algorithm to solve the optimized DOA estimation model, to realize the reconstruction of the original signal, and obtain the reconstructed original signal.

[0138] In specific implementation, after obtaining the optimized DOA estimation model, the non-convex optimization algorithm design can be performed, and the momentum gradient descent algorithm is used to solve the optimized DOA estimation model (iterative update) to realize the reconstruction of the original signal, thereby obtaining the reconstructed original signal.

[0139] In the reconstruction process, first, based on the traditional spectral initialization method, the normalized spectral initialization processing of and is needed to obtain the initialized and Subsequently, a momentum gradient descent algorithm was designed by combining the momentum acceleration method, and the momentum gradient descent algorithm was used based on... and The optimized DOA estimation model was solved, and a total of [number] steps were performed. Round iteration ( (Integers greater than 1). Based on... The output results after each iteration are used to reconstruct the original signal, resulting in the reconstructed original signal.

[0140] In progress and During the spectral initialization process, the following formula is first used to initialize the spectrum. Upgraded Hankel matrix Perform truncated singular value decomposition:

[0141] ;

[0142] in, express The rank approximation, , , .

[0143] Then, based on the following formula, , , right and Perform normalized spectral initialization to obtain the initialized spectrum. and :

[0144] ;

[0145] ;

[0146] in, This represents the Frobenius norm normalization operator.

[0147] After completing the and After spectral initialization, the momentum gradient descent algorithm can be designed using the momentum acceleration method. The following formula utilizes the momentum gradient descent algorithm based on... and Solve the optimized DOA estimation model:

[0148] ;

[0149] ;

[0150] ;

[0151] ;

[0152] wherein, denotes a momentum parameter, denotes a step size, denotes the i-th iteration, denotes the i-th iteration, and denote the gradient in the direction of the i-th iteration and the gradient in the direction of the i-th iteration, respectively. Finally, based on the output results after i iterations, the reconstruction of the original signal is performed to obtain the reconstructed original signal

[0153] which is expressed by the formula:

[0154] .

[0155] In one example, let .

[0156] When , the update of and is performed by the following formula:

[0157] ;

[0158] .

[0159] When , the update of and is performed by the following formula:

[0160] ;

[0161] .

[0162] Step 15, the reconstructed original signal is processed by using the single-snapshot subspace method to finally obtain the estimated value of the direction of arrival of each signal source.

[0163] In the specific implementation, after obtaining the reconstructed original signal, the reconstructed original signal can be processed by using the single-snapshot subspace method to finally obtain the estimated value of the direction of arrival of each signal source .

[0164] ​​​​​​The single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition provided in the embodiment brings the following beneficial effects.

[0165] First, the hardware complexity and power consumption are reduced. The single-bit sampling method is adopted in the embodiment, only the signal symbol information is retained to complete the processing, and the dependence on high-precision analog-to-digital converters (ADCs) is avoided, thereby significantly reducing the hardware cost and system energy consumption.

[0166] Second, the single-snapshot observation condition is adapted. The structural characteristics of the single-snapshot signal are fully tapped by constructing the Hankel matrix and introducing the low-rank constraint, and the problem that the traditional method is difficult to realize stable direction of arrival estimation under the single-snapshot condition is effectively solved.

[0167] Third, the precision and robustness of the direction of arrival estimation are improved. The sign consistency constraint is introduced in the model construction, and the Hankel matrix decomposition and the function smoothing method are combined for optimization processing, which can ensure the consistency between the reconstructed signal and the observed signal, thereby maintaining high estimation precision and robustness in a low signal-to-noise ratio environment.

[0168] Fourth, the calculation efficiency is improved, and real-time processing is facilitated. The momentum gradient descent algorithm is used to solve the optimization model, which can accelerate the convergence speed and avoid falling into local optimum, and has lower computational complexity than the traditional convex optimization method, and is more suitable for real-time signal processing scenarios.

[0169] The step division of the above methods is only for the purpose of clear description, and in implementation, one step can be combined or some steps can be divided into multiple steps, as long as the same logical relationship is included, and all are within the protection scope of the present application. Irrelevant modifications or irrelevant designs are added to the algorithm or the flow, but the core design of the algorithm and the flow is not changed, and all are within the protection scope of the present application.

[0170] In one embodiment, in order to verify the effectiveness of the single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition (hereinafter referred to as the method or Acc-OSCAR) proposed in the present application, we carried out the corresponding simulation experiment.

[0171] The experiment shows the performance of the method (Acc-OSCAR- and Acc-OSCAR- ) in single-bit direction of arrival estimation, and compares it with the advanced algorithms CBIHT- , CBIHT- , iCBIHT and 1bit-ANM. Direction of arrival estimation for a uniform linear array of sensors, setting... , The vector consisting of the direction of arrival and the direction of arrival .

[0172] Figure 4 The success rates of this method and the comparison algorithm are shown. Figure 4 As can be seen from this, this method has the best success rate compared to 1-bit-ANM. The success rate represents... The RMSE of the sub-Monte Carlo algorithm is less than the threshold. The frequency.

[0173] Figure 5 This demonstrates the RMSE of our method and the comparison algorithm under successful conditions, from Figure 5 It can be seen that this method has better RMSE than 1bit-ANM, especially when the signal-to-noise ratio is greater than 20dB.

[0174] Another embodiment of this application proposes a single-bit single-snapshot signal direction-of-arrival estimation system based on Hankel matrix decomposition. The details of the single-bit single-snapshot signal direction-of-arrival estimation system based on Hankel matrix decomposition proposed in this embodiment are described in detail below. The following content is only for the convenience of understanding and is not necessary for implementing this example.

[0175] Figure 6 This is a schematic diagram of the structure of a single-bit single-shot signal direction-of-arrival estimation system based on Hankel matrix decomposition proposed in this embodiment, including: acquisition module 21, construction module 22, optimization module 23, original signal reconstruction module 24, and estimation execution module 25.

[0176] The acquisition module 21 is used to acquire the single-shot sampling signal received by the vehicle-mounted millimeter-wave radar array, as well as the single-bit quantization signal corresponding to the single-shot sampling signal.

[0177] Module 22 is used to construct an initial DOA estimation model based on the symbol consistency constraint between the single-snapshot sampled signal and the single-bit quantized signal and the Hankel low-rank constraint of the single-snapshot sampled signal.

[0178] Optimization module 23 is used to optimize the initial DOA estimation model using Hankel matrix decomposition and function smoothing methods to obtain the optimized DOA estimation model.

[0179] The original signal reconstruction module 24 is used to solve the optimized DOA estimation model using the momentum gradient descent algorithm to reconstruct the original signal and obtain the reconstructed original signal.

[0180] An estimation executing module 25 is configured to process the reconstructed original signals by using the single snapshot subspace method, and finally obtain the estimation of the direction of arrival of each signal source.

[0181] It can be found that the embodiment is a system embodiment corresponding to the method embodiment, and the embodiment can be implemented in cooperation with the method embodiment. The related technical details and technical effects mentioned in the method embodiment are still valid in the embodiment. In order to reduce repetition, they will not be described here. Correspondingly, the related technical details mentioned in the embodiment can also be applied to the method embodiment.

[0182] It is worth mentioning that each module and module involved in the embodiment is a logical module. In actual application, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of the present application, units not closely related to solving the technical problems proposed in the present application are not introduced in the embodiment, but this does not mean that there are no other units in the embodiment.

[0183] Another embodiment of the present application provides an electronic device, as shown in the figure, comprising a processor 31 and a memory 32, the memory 32 stores instructions executable by the processor 31, and the processor 31 is configured to execute the instructions, so that the electronic device can implement a single-bit single-shot signal direction of arrival estimation method based on Hankel matrix decomposition as described in the method embodiment. Figure 7

[0184] Wherein, the memory and the processor are connected in a bus mode, the bus includes any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripheral devices, voltage stabilizers and power management circuits together, which are well known in the art, so this document will not further describe them. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements such as multiple receivers and transmitters, which provide a unit for communicating with various other devices on the transmission medium. The data processed by the processor is transmitted on the wireless medium through the antenna, and further, the antenna also receives data and transmits the data to the processor.

[0185] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interface, voltage regulation, power management and other control functions. And the memory can be used to store the data used by the processor in the execution operation.

[0186] ​Another embodiment of the present application provides a computer readable storage medium, wherein a computer program is stored in the computer readable storage medium, and the computer program, when executed by a processor, enables a single-bit single-shot signal direction of arrival estimation method based on Hankel matrix decomposition to be implemented.

[0187] That is, a person skilled in the art can understand that all or part of the steps in the above method embodiments can be completed by programs instructing relevant hardware, and the programs are stored in a storage medium and include a plurality of instructions for enabling a device (such as a single-chip microcomputer, a chip, etc.) or a processor to execute all or part of the steps of the method described in the method embodiments of the present application. The foregoing storage medium includes a U disk, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk, and various media that can store program codes.

[0188] A person skilled in the art can understand that each of the above embodiments is a specific embodiment for implementing the present application, and in actual application, various changes can be made in form and details without departing from the spirit and scope of the present application. For those skilled in the art, a number of improvements and refinements can be made without departing from the principles of the present application, and these improvements and refinements are also considered to be within the protection scope of the present application.

Claims

1. A single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition, characterized in that, The method comprises: acquiring a single-shot sampling signal received by a vehicle-mounted millimeter wave radar array and a single-bit quantized signal corresponding to the single-shot sampling signal; based on the sign consistency constraint between the single-shot sampling signal and the single-bit quantized signal and the Hankel low-rank constraint of the single-shot sampling signal, an initial DOA estimation model is constructed; using a Hankel matrix decomposition and function smoothing method, the initial DOA estimation model is optimized to obtain an optimized DOA estimation model; using a momentum gradient descent algorithm, the optimized DOA estimation model is solved to realize reconstruction of the original signal and obtain a reconstructed original signal; using a single-shot subspace method, the reconstructed original signal is processed to finally obtain an estimated value of the direction of arrival of each signal source; The vehicle-mounted millimeter wave radar array is a uniform linear array, which is composed of a total of array elements, is an integer greater than 1, and the array element spacing is , the single-snapshot sampling signal received by the vehicle-mounted millimeter wave radar array is represented by the formula as: ； ； ； wherein denotes a noise vector, , denotes a vector space of dimension , the upper right index denotes the transpose operation, denotes the signal received by the th array element of the vehicle-mounted millimeter wave radar array, denotes the spectral sparsity, i.e. the number of signal sources, denotes the amplitude of the echo of the th signal source, denotes the wavelength of the echo, denotes the direction of arrival of the th signal source, i.e. the direction of incidence of the th signal source; The note single snapshot spectrum sparse signal is , = , then Based on is expressed as: ； Based on , the corresponding single-bit quantization signal of , , is expressed by the formula: ； wherein, represents taking the real part, represents taking the imaginary part, represents taking the sign element-wise; the sign consistency constraint between the single-shot sampling signal and the single-bit quantized signal is expressed by a formula as: ； wherein represents the conjugate of represents a dot product operation; Let denote the upsize operator of Hankel matrix, for upsize to , , denote the dimension of the vector space after upsize, then the Hankel low-rankness constraint of single-shot sampling signal is represented by the formula: ； wherein denotes a lifted Hankel matrix of denotes the row of a Hankel matrix, denotes the column of a Hankel matrix; based on the sign consistency constraint between the single-shot sampling signal and the single-bit quantized signal and the Hankel low-rank constraint of the single-shot sampling signal, an initial DOA estimation model is constructed, including: By utilizing the sign consistency constraint between the single-shot sampled signal and the single-bit quantized signal, the signal to be reconstructed... symbols and To maintain sign consistency, the Hankel low-rank constraint of the single-shot sampled signal is utilized. Upgraded Hankel matrix rank and The rank remains consistent; Based on , an initial DOA estimation model is constructed, and the initial DOA estimation model is represented as: ； wherein, denotes the one-sided norm, upper right superscript 2 denotes squaring. denotes taking the power of, denotes retaining elements with negative absolute values and zeroing elements with positive absolute values, is a positive integer, in the case of sparse noise including impulse noise, takes , in the case of dense noise including white noise, takes , denotes norm, upper right superscript 2 denotes squaring.

2. The single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition according to claim 1, characterized in that, using a Hankel matrix decomposition and function smoothing method, the initial DOA estimation model is optimized to obtain an optimized DOA estimation model, including: The Hankel matrix decomposition is used to decompose into , where , and the upper right corner represents the conjugate transpose operation, and are iteration variables, , ; To ensure with Hankel low-rank structure, adding operator constraints, the operator constraints are represented as: ； ； wherein denotes the identity operator, denotes the linear operator, denotes the inverse operator of denotes the adjoint operator of based on the operator constraint, the initial DOA estimation model is converted into an intermediate DOA estimation model, which is expressed as: ； The function smoothing method is used to smooth to , for the variable , satisfies: ； wherein is a smoothing parameter, ; based on the intermediate DOA estimation model, the Hankel low-rank constraint of the single-shot sampling signal is converted into a penalty term to obtain an optimized DOA estimation model, which is expressed as: ； wherein is a penalty parameter, , denotes an identity relation.

3. The method of claim 2, wherein, using a momentum gradient descent algorithm, the optimized DOA estimation model is solved to realize reconstruction of the original signal and obtain a reconstructed original signal, including: To and normalization, the spectrum initialization processing is performed to obtain the initialized and ; The momentum gradient descent algorithm is designed in combination with the momentum acceleration method, and the momentum gradient descent algorithm is used to solve the optimized DOA estimation model and The optimized DOA estimation model is solved, and a total of rounds of iterations are performed, is an integer greater than 1. Based on The output results of the iterations are used to reconstruct the original signal to obtain a reconstructed original signal.

4. The single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition according to claim 3, characterized in that, To and a spectrum initialization process that normalizes the spectrum to obtain an initialized and , comprising: The truncated singular value decomposition of the lifted Hankel matrix of order is performed by the following equation: ； wherein denotes rank approximation of , , ; The spectral initialization processing is performed on the basis of , , and by the following equation to obtain initialized and :​ ； ； wherein denotes the Frobenius norm normalization operator; The momentum gradient descent algorithm is designed in combination with the momentum acceleration method, and the momentum gradient descent algorithm is used to solve the optimized DOA estimation model based on and The optimized DOA estimation model is solved, and the following formula is used to achieve this: ； ； ； ； wherein represents a momentum parameter, represents a step size, represents the i-th iteration, and represent the gradient taken in the direction of and the gradient taken in the direction of respectively, Based on the output result after iteration of the wheel and reconstruct the original signal to obtain the reconstructed original signal , is expressed by the formula: 。 5. The method of claim 4, wherein, Set ; When , update and by the following equations: ； ； When , update and by the following equations: ； 。 6. A one-bit one-shot signal direction of arrival estimation system based on Hankel matrix decomposition, implemented based on a one-bit one-shot signal direction of arrival estimation method based on Hankel matrix decomposition according to any one of claims 1 to 5, characterized in that, The system comprises: an acquisition module configured to acquire a single-shot sampling signal received by a vehicle-mounted millimeter wave radar array and a single-bit quantized signal corresponding to the single-shot sampling signal; a construction module configured to construct an initial DOA estimation model based on the sign consistency constraint between the single-shot sampling signal and the single-bit quantized signal and the Hankel low-rank constraint of the single-shot sampling signal; an optimization module configured to optimize the initial DOA estimation model using a Hankel matrix decomposition and function smoothing method to obtain an optimized DOA estimation model; an original signal reconstruction module configured to solve the optimized DOA estimation model using a momentum gradient descent algorithm to realize reconstruction of the original signal and obtain a reconstructed original signal; an estimation execution module configured to process the reconstructed original signal using a single-shot subspace method to finally obtain an estimated value of the direction of arrival of each signal source.

7. An electronic device, comprising: including: a processor and a memory, the memory storing instructions executable by the processor, and the processor being configured to execute the instructions to enable the electronic device to implement a single-bit single-shot signal direction of arrival estimation method based on Hankel matrix decomposition according to any one of claims 1 to 5.

8. A computer readable storage medium storing a computer program, characterized in that, The computer program is executed by a processor to implement the single-bit single-snapshot signal direction of arrival estimation method based on Hankel matrix decomposition in any one of claims 1 to 5.

Images