Block sparse adaptive filter used in non-Gaussian noise environment

Abstract

The invention provides a block sparse adaptive filter used in a non-Gaussian noise environment. The filter adopts an adaptive filtering algorithm of a mixed norm and an arc tangent minimum mean square error to update a filter vector so as to obtain an actual optimal value of the filter vector; the mixed norm is obtained based on the filter vector after uniform blocking; the updating step comprises the steps of setting an optimal value based on an input vector of a filter and a filter vector to obtain a target output vector; a filter vector updating formula is determined based on an arc tangent minimum mean square error self-adaptive filtering algorithm, the filter vector updating formula comprises a proportional adjustment matrix, and the proportional adjustment matrix is obtained through mixed norm calculation; obtaining a current output vector based on the input vector of the filter and the current filter vector; and if the difference value between the current output vector and the target output vector does not meet the iteration termination requirement, updating the current filter vector by using a filter vector updating formula until the iteration termination requirement is met, and outputting the actual optimal value of the filter vector.

Description

Technical Field

[0001] The present invention relates to the field of adaptive filtering technology, and in particular to a block sparse adaptive filter used in a non-Gaussian noise environment. Background Art

[0002] In recent years, with the development of digital signal processing technology, adaptive filters have been widely used in the fields of channel equalization, channel estimation, echo cancellation, and interference cancellation. Among them, the least mean square algorithm (LMS) has the advantages of simple implementation, good performance, and high robustness, and is the most widely used algorithm in adaptive filtering. However, in many practical application scenarios, the system is sparse or block sparse, which means that most of the energy is distributed on only a few taps, and the other taps are zero or approximately zero. In addition, many systems are affected by non-Gaussian noise, such as underwater acoustic communication systems. In the above case, the performance of the LMS algorithm is significantly reduced. Therefore, it is necessary to design a block sparse adaptive filter suitable for non-Gaussian noise environment based on the characteristics of the actual system to improve the convergence speed and accuracy of the system.

[0003] Sparse adaptive filter designs fall into two categories. One is the norm-regularized filtering algorithm, which explicitly adds sparsity-inducing regularization terms (such as L1, L0, and log-sum penalties) to the cost function, directly constraining the sparsity of the solution through the optimization process. This type of algorithm has high computational complexity, and the regularization term slows down the convergence of large coefficient taps. The other is the proportional adjustment filtering algorithm, which dynamically adjusts the filter coefficient update step size to achieve faster convergence for larger coefficients (the key component that dominates sparsity) while suppressing the update amplitude of small coefficients. This type of algorithm only has good processing performance in sparse systems. Summary of the Invention

[0004] The present invention aims to solve one of the technical problems in the related art at least to a certain extent.

[0005] To this end, a first object of the present invention is to propose a block sparse adaptive filter for use in a non-Gaussian noise environment to improve the convergence speed of the filter.

[0006] The second object of the present invention is to propose a block sparse adaptive filter optimization method for non-Gaussian noise environment.

[0007] To achieve the above-mentioned object, the first aspect of the present invention proposes a block sparse adaptive filter for use in a non-Gaussian noise environment, wherein the filter uses a hybrid norm and an inverse tangent minimum mean square error adaptive filtering algorithm to update a filter vector to obtain an actual optimal value of the filter vector; the hybrid norm is obtained based on the filter vector after uniform block division; and the step of updating the filter vector comprises:

[0008] Setting an optimal value based on the input vector of the filter and the filter vector to obtain a target output vector;

[0009] Determining a filter vector update formula based on an adaptive filtering algorithm of arctangent minimum mean square error, wherein the filter vector update formula includes a proportional adjustment matrix, and the proportional adjustment matrix is calculated by the mixed norm;

[0010] Obtaining a current output vector based on an input vector of the filter and a current filter vector;

[0011] If the difference between the current output vector and the target output vector does not meet the iteration termination requirement, the current filter vector is updated using the filter vector update formula until the iteration termination requirement is met, and the actual optimal value of the filter vector is output.

[0012] In the block sparse adaptive filter for non-Gaussian noise environment provided by the first aspect of the present invention, the process of obtaining the mixed norm includes: calculating the 1 norm for each vector block in the filter vector after uniform blocking, and then calculating the 0 norm, so as to obtain the mixed norm.

[0013] In the block sparse adaptive filter for non-Gaussian noise environment provided by the first aspect of the present invention, if the length of the last vector block is insufficient when the filter vector is evenly divided into blocks, a zero padding operation is performed.

[0014] In the block sparse adaptive filter for non-Gaussian noise environment provided by the first aspect of the present invention, the adaptive filtering algorithm based on the inverse tangent minimum mean square error determines the filter vector update formula, including: obtaining a cost function based on the adaptive filtering algorithm based on the inverse tangent minimum mean square error; determining the filter vector update formula based on the cost function and the proportional adjustment matrix.

[0015] In the block sparse adaptive filter for a non-Gaussian noise environment provided in the first aspect of the present invention, the filter vector update formula is determined based on the cost function and the proportional adjustment matrix, including: minimizing the cost function using stochastic gradient descent, and determining the filter vector update formula in combination with the proportional adjustment matrix.

[0016] In the block sparse adaptive filter for non-Gaussian noise environment provided by the first aspect of the present invention, the iteration termination requirement means that the number of iterations reaches the maximum number of iterations or the difference is less than the set difference threshold.

[0017] To achieve the above-mentioned object, the second aspect of the present invention proposes a block sparse adaptive filter optimization method for a non-Gaussian noise environment, which is applicable to the block sparse adaptive filter for a non-Gaussian noise environment provided by the first aspect of the present invention, comprising:

[0018] Obtaining a mixed norm based on the filter vector after uniform block division;

[0019] Setting an optimal value based on the input vector of the filter and the filter vector to obtain a target output vector;

[0020] Determining a filter vector update formula based on an adaptive filtering algorithm of arctangent minimum mean square error, wherein the filter vector update formula includes a proportional adjustment matrix, and the proportional adjustment matrix is calculated by the mixed norm;

[0021] Obtaining a current output vector based on an input vector of the filter and a current filter vector;

[0022] If the difference between the current output vector and the target output vector does not meet the iteration termination requirement, the current filter vector is updated using the filter vector update formula until the iteration termination requirement is met, and the actual optimal value of the filter vector is output to update the filter vector, thereby obtaining the optimal block sparse adaptive filter.

[0023] In the block sparse adaptive filter optimization method for a non-Gaussian noise environment provided in the second aspect of the present invention, the mixed norm is obtained based on the filter vector after uniform blocking, including: calculating the 1 norm for each vector block in the filter vector after uniform blocking, and then calculating the 0 norm, thereby obtaining the mixed norm.

[0024] In the block sparse adaptive filter optimization method for a non-Gaussian noise environment provided in the second aspect of the present invention, the adaptive filtering algorithm based on the inverse tangent minimum mean square error determines the filter vector update formula, including: obtaining a cost function based on the adaptive filtering algorithm based on the inverse tangent minimum mean square error; determining the filter vector update formula based on the cost function and the proportional adjustment matrix.

[0025] In the block sparse adaptive filter optimization method for a non-Gaussian noise environment provided in the second aspect of the present invention, the filter vector update formula is determined based on the cost function and the proportional adjustment matrix, including: minimizing the cost function using stochastic gradient descent, and determining the filter vector update formula in combination with the proportional adjustment matrix.

[0026] The present invention provides a block sparse adaptive filter for use in a non-Gaussian noise environment, wherein the filter uses a hybrid norm and an inverse tangent minimum mean square error adaptive filtering algorithm to update a filter vector to obtain an actual optimal value of the filter vector; the hybrid norm is obtained based on the filter vector after uniform block division; the step of updating the filter vector includes: setting an optimal value based on the filter input vector and the filter vector to obtain a target output vector; determining a filter vector update formula based on the adaptive filtering algorithm of the inverse tangent minimum mean square error, the filter vector update formula includes a proportional adjustment matrix, and the proportional adjustment matrix is calculated by the hybrid norm; obtaining a current output vector based on the filter input vector and the current filter vector; if the difference between the current output vector and the target output vector does not meet the iteration termination requirement, then using the filter vector update formula to update the current filter vector until the iteration termination requirement is met, and outputting the actual optimal value of the filter vector. In this case, the filter uses a hybrid norm and an inverse tangent minimum mean square error adaptive filtering algorithm to update the filter vector; wherein the hybrid norm is obtained based on the filter vector after uniform block division, and in the step of updating the filter vector, the filter vector update formula includes a proportional adjustment matrix, and the proportional adjustment matrix is calculated by the hybrid norm. Therefore, the hybrid norm can be used as a metric to allocate the step size of the filter taps, so that blocks with more active taps are allocated larger step sizes, thereby accelerating the convergence speed of the filter.

[0027] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0029] Figure 1 An update block diagram of a block sparse adaptive filter for a non-Gaussian noise environment provided by an embodiment of the present invention;

[0030] Figure 2 This is a flowchart of a block sparse adaptive filter optimization method for a non-Gaussian noise environment provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible implementations consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with certain aspects of the present invention, as detailed in the appended claims.

[0032] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0033] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. It should also be understood that the term "and / or" used in the present invention refers to and includes any or all possible combinations of one or more associated listed items.

[0034] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0035] The present invention proposes a block sparse adaptive filter for use in a non-Gaussian noise environment to improve the convergence speed of the filter. The block sparse adaptive filter for use in a non-Gaussian noise environment in the present invention can be simply referred to as a filter.

[0036] In an embodiment of the present invention, a block sparse adaptive filter for a non-Gaussian noise environment uses a hybrid norm and arctangent minimum mean square error adaptive filtering algorithm to update the filter vector to obtain the actual optimal value of the filter vector. The hybrid norm is obtained based on the filter vector after uniform block division.

[0037] Figure 1 This is a block diagram of an update of a block sparse adaptive filter for a non-Gaussian noise environment provided by an embodiment of the present invention.

[0038] In an embodiment of the present invention, the process of updating the filter vector includes: determining a filter vector update formula based on an adaptive filtering algorithm of arctangent minimum mean square error, the filter vector update formula including a proportional adjustment matrix, the proportional adjustment matrix being calculated by a mixed norm; setting an optimal value based on the input vector of the filter and the filter vector to obtain a target output vector; obtaining a current output vector based on the input vector of the filter and the current filter vector of the block sparse adaptive filter for a non-Gaussian noise environment; calculating the difference between the current output vector and the target output vector, and if the difference between the current output vector and the target output vector does not meet the iteration termination requirement, updating the current filter vector using the filter vector update formula until the iteration termination requirement is met, and outputting the actual optimal value of the filter vector (see Figure 1 ).

[0039] In an embodiment of the present invention, the process of obtaining the mixed norm includes: calculating the 1-norm for each vector block (i.e., each block) in the uniformly partitioned filter vector, and then calculating the 0-norm, thereby obtaining the mixed norm. If the length of the last vector block is insufficient when the filter vector is uniformly partitioned, a zero-padding operation is performed.

[0040] Specifically, the present invention adopts a uniform block processing method to divide the system to be processed (i.e., the block sparse adaptive filter for non-Gaussian noise environment of the present invention) into blocks of a certain length. Assume that the block length is B, and the system to be processed with a length of L is divided into N blocks, that is, L = B*N. In some embodiments, if the length L of the adaptive filter cannot be divided by B, zero extension can be added to the end of the filter vector. The filter vector can be expressed as:

[0041]

[0042] Where w is the filter vector. L is the length of the filter vector (i.e., the total number of filter taps). The value of the i-th filter tap is w i , i ranges from 1 to L. [] T Represents the transpose operation. N is the number of vector blocks divided by the filter vector. The N vector blocks are w [1] 、…、w [k] 、…、w [N] The kth vector block can be represented as w [k] k ranges from 1 to N. B is the length of each vector block (i.e., the block length).

[0043] Among them, the mixed norm can be expressed as:

[0044] ||w|| 1，0 =||[||w [1] ||1,||w [2] ||1,…,||w[N] ||1] T ||0

[0045] Where, ||w|| 1,0 represents a mixed norm. ||·||1 represents the calculation of the 1-norm. ||·||0 represents the calculation of the 0-norm. Since the 0-norm is an NP-hard problem (i.e., a non-deterministic problem of polynomial complexity), the following formula is used for approximation:

[0046]

[0047] Wherein, the value of parameter β is preset.

[0048] In an embodiment of the present invention, determining a filter vector update formula based on an adaptive filtering algorithm using an inverse tangent minimum mean square error includes: obtaining a cost function based on the adaptive filtering algorithm using an inverse tangent minimum mean square error; and determining the filter vector update formula based on the cost function and a proportional adjustment matrix. Determining the filter vector update formula based on the cost function and the proportional adjustment matrix includes: minimizing the cost function using stochastic gradient descent and determining the filter vector update formula in combination with the proportional adjustment matrix.

[0049] Specifically, considering the inverse tangent framework, the cost function of the adaptive filtering algorithm based on the inverse tangent minimum mean square error is expressed as:

[0050]

[0051] Where, J K is the cost function. γ is the control Introducing a positive parameter of size, E[] represents the mean. e K is the error distribution.

[0052] Minimize the cost function J using stochastic gradient descent K , we get the initial iterative expression, and then introduce the proportional adjustment matrix G(n) calculated by the block mixed norm. Therefore, the filter vector iterative expression (i.e., update formula) obtained by the block adaptive filtering algorithm based on the inverse tangent minimum mean square error is:

[0053] w(n+1)=w(n)+μf(e(n))G(n)x(n)

[0054] Where w(n+1) is the filter vector at time n+1, w(n) is the filter vector at time n, and μ is the step size factor. This parameter plays an important role in reducing steady-state misalignment. As can be seen, when the value of e(n) is large, the function f(e(n)) becomes zero, which improves the algorithm's robustness to impulse noise or outliers. The parameter γ controls the steepness of the cost function. When γ → 0, the algorithm behaves like the LMS algorithm. When γ → ∞ and f(e(n)) → 0, the convergence rate decreases.

[0055] The proportional adjustment matrix G(n) is:

[0056] G(n)=diag{g1(n)1 B , g2(n)1 B ,...,g N (n)1 B}

[0057] Where, 1 B Represents a vector of all 1s of length B. m ranges from 1 to N.

[0058] When m is j, we can calculate The parameter ρ prevents w(n) from affecting processing performance when it is much smaller than the maximum coefficient. w [m](n) represents the mth vector block at time n. When all coefficients of w(n) are zero, the parameter q is used for regularization update. The value of parameter q can be preset.

[0059] In the embodiment of the present invention, the iteration termination requirement refers to the number of iterations reaching the maximum number of iterations or the difference being less than a set difference threshold.

[0060] In an embodiment of the present invention, considering the identification problem of the system to be processed, the input vector of the system to be processed, i.e., the filter, is expressed as:

[0061] x(n)=[x(n),x(n-1),...,x(n-L+1)]T

[0062] Where x(n) is the input vector at time n. The length of the input vector is equal to L.

[0063] The optimal value of the filter vector of the system to be processed is defined as w0, and the target output vector of the system to be processed is expressed as:

[0064] d(n)=x(n) T w0+v(n)

[0065] Where d(n) is the target output vector. v(n) is the background noise with zero mean. The current filter vector of the system to be processed is defined as: w(n) = [w1(n), w2(n), ..., w L(n)] T , then the current output vector of the system to be processed is: x(n) T w(n). Calculate the difference between the current output vector and the target output vector, that is, at time n, the measurement error of the adaptive filter is:

[0066] e(n)=d(n)-x(n) T w(n)

[0067] Where e(n) is the difference between the previous output vector and the target output vector. The distribution of e(n) is the error distribution e K .

[0068] In an embodiment of the present invention, if the difference between the current output vector and the target output vector is less than a set difference threshold or the number of iterations reaches a maximum number of iterations, the actual optimal value of the filter vector (i.e., the current filter vector at the last iteration) is output. If the iteration termination requirement is not met, the current filter vector is updated using a filter vector update formula until the iteration termination requirement is met. The actual optimal value of the filter vector is assigned to the block sparse adaptive filter for non-Gaussian noise environment of the present invention to obtain the desired block sparse adaptive filter.

[0069] This paper introduces a mixed norm into the proportional adjustment matrix, where the 1-norm represents sparsity within a block, and the 0-norm constrains sparsity between blocks. During the iterative process, the proportional adjustment concept is employed, using the mixed norm of the estimated filter coefficients within a block as a metric to allocate the step size of the filter taps. This results in blocks with more active taps being assigned larger step sizes, thereby accelerating the algorithm's convergence. This algorithm achieves fast convergence, good tracking, and strong stability in non-Gaussian noise environments.

[0070] The following is a method embodiment of the present invention. For details not disclosed in the method embodiment of the present invention, please refer to the filter embodiment of the present invention. The method embodiment of the present invention proposes a method for optimizing a block sparse adaptive filter for use in a non-Gaussian noise environment. This method for optimizing a block sparse adaptive filter for use in a non-Gaussian noise environment utilizes the block sparse adaptive filter for use in a non-Gaussian noise environment of the filter embodiment described above. The method for optimizing a block sparse adaptive filter for use in a non-Gaussian noise environment of the present invention can be simply referred to as an optimization method.

[0071] Figure 2 This is a flowchart of a block sparse adaptive filter optimization method for a non-Gaussian noise environment provided by an embodiment of the present invention.

[0072] like Figure 2 As shown, the block sparse adaptive filter optimization method for non-Gaussian noise environment includes:

[0073] Step S101, obtaining a mixed norm based on the filter vector after uniform blocking;

[0074] Step S102, obtaining a target output vector based on the filter input vector and the filter vector setting optimal value;

[0075] Step S103, determining a filter vector update formula based on an inverse tangent minimum mean square error adaptive filtering algorithm, where the filter vector update formula includes a proportional adjustment matrix, which is calculated using a mixed norm;

[0076] Step S104, obtaining a current output vector based on the filter input vector and the current filter vector;

[0077] Step S105: If the difference between the current output vector and the target output vector does not meet the iteration termination requirement, the current filter vector is updated using the filter vector update formula until the iteration termination requirement is met, and the actual optimal value of the filter vector is output to update the filter vector, thereby obtaining the optimal block sparse adaptive filter.

[0078] In step S101, a mixed norm is obtained based on the uniformly partitioned filter vector, including: calculating a 1-norm for each vector block in the uniformly partitioned filter vector, and then calculating a 0-norm, thereby obtaining the mixed norm. If the last vector block is not long enough when the filter vector is uniformly partitioned, zero padding is performed.

[0079] In step S103, a filter vector update formula is determined based on an adaptive filtering algorithm of inverse tangent minimum mean square error, including: obtaining a cost function based on an adaptive filtering algorithm of inverse tangent minimum mean square error; and determining a filter vector update formula based on the cost function and a proportional adjustment matrix.

[0080] In step S103 , a filter vector update formula is determined based on the cost function and the proportional adjustment matrix, including: minimizing the cost function by using stochastic gradient descent, and determining the filter vector update formula in combination with the proportional adjustment matrix.

[0081] In step S105 , the iteration termination requirement refers to the number of iterations reaching the maximum number of iterations or the difference being less than a set difference threshold.

[0082] It should be noted that the above explanation of the embodiment of the block sparse adaptive filter for non-Gaussian noise environment is also applicable to the optimization method of the block sparse adaptive filter for non-Gaussian noise environment of this embodiment, which will not be repeated here.

[0083] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0084] In a block sparse adaptive filter for a non-Gaussian noise environment according to an embodiment of the present invention, the filter uses a hybrid norm and an inverse tangent minimum mean square error adaptive filtering algorithm to update a filter vector to obtain an actual optimal value of the filter vector; the hybrid norm is obtained based on the filter vector after uniform block division; the step of updating the filter vector includes: setting an optimal value based on the filter input vector and the filter vector to obtain a target output vector; determining a filter vector update formula based on the adaptive filtering algorithm of the inverse tangent minimum mean square error, the filter vector update formula includes a proportional adjustment matrix, and the proportional adjustment matrix is calculated by the hybrid norm; obtaining a current output vector based on the filter input vector and the current filter vector; if the difference between the current output vector and the target output vector does not meet the iteration termination requirement, the filter vector update formula is used to update the current filter vector until the iteration termination requirement is met, and the actual optimal value of the filter vector is output. In this case, the filter uses a hybrid norm and an inverse tangent minimum mean square error adaptive filtering algorithm to update the filter vector; wherein the hybrid norm is obtained based on the filter vector after uniform block division, and in the step of updating the filter vector, the filter vector update formula includes a proportional adjustment matrix, and the proportional adjustment matrix is calculated by the hybrid norm. Therefore, the hybrid norm can be used as a metric to allocate the step size of the filter taps, so that blocks with more active taps are allocated larger step sizes, thereby accelerating the convergence speed of the filter.

[0085] The filter and optimization method of the present invention take into account the block sparsity characteristics of the system and add a matrix G(n) with a mixed norm constraint to the filter iteration expression to accelerate the convergence speed of the algorithm. This matrix is a diagonal matrix, the elements of which are determined by the sparsity of the taps in each block. When a block contains more important taps, the mixed norm of the block is a larger value, which increases the proportional factor corresponding to the matrix G(n), thereby accelerating the convergence speed of the filter coefficients in the block. In addition, compared with the algorithm that only updates one tap in each iteration, the algorithm proposed in the present invention can use the mixed norm to allocate step sizes for multiple taps in a block, reducing the amount of calculation and improving the convergence speed of the algorithm. This method can greatly improve the convergence speed of the block sparse adaptive filter, which is of great significance for achieving accurate estimation of block sparse systems in non-Gaussian noise environments.

[0086] The accompanying drawings illustrate schematic diagrams of the structures of the disclosed embodiments of the present invention. These figures are not drawn to scale; for the purpose of clarity, some details are exaggerated and some details may be omitted. The shapes of the various regions and layers shown in the figures, as well as their relative sizes and positional relationships, are merely exemplary and may deviate in practice due to manufacturing tolerances or technical limitations. Those skilled in the art may design regions / layers with different shapes, sizes, and relative positions as needed.

[0087] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in the present invention can be achieved. The present invention is not limited here.

[0088] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A block sparse adaptive filter for use in a non-Gaussian noise environment, characterized in that: The filter uses a hybrid norm and an inverse tangent minimum mean square error adaptive filtering algorithm to update a filter vector to obtain an actual optimal value of the filter vector; the hybrid norm is obtained based on the filter vector after uniform block division; the step of updating the filter vector includes: Setting an optimal value based on the input vector of the filter and the filter vector to obtain a target output vector; Determining a filter vector update formula based on an adaptive filtering algorithm of arctangent minimum mean square error, wherein the filter vector update formula includes a proportional adjustment matrix, and the proportional adjustment matrix is calculated by the mixed norm; Obtaining a current output vector based on an input vector of the filter and a current filter vector; If the difference between the current output vector and the target output vector does not meet the iteration termination requirement, the current filter vector is updated using the filter vector update formula until the iteration termination requirement is met, and the actual optimal value of the filter vector is output.

2. The block sparse adaptive filter for non-Gaussian noise environment according to claim 1, characterized in that The process of obtaining the mixed norm includes: The 1-norm is calculated for each vector block in the filter vector after uniform blocking, and then the 0-norm is calculated, thereby obtaining the mixed norm.

3. The block sparse adaptive filter for non-Gaussian noise environment according to claim 1, characterized in that When the filter vector is evenly divided into blocks, if the length of the last vector block is insufficient, a zero padding operation is performed.

4. The block sparse adaptive filter for non-Gaussian noise environment according to claim 1, characterized in that The adaptive filtering algorithm based on inverse tangent minimum mean square error determines the filter vector update formula, including: The cost function is obtained based on the adaptive filtering algorithm of arctangent minimum mean square error; A filter vector update formula is determined based on the cost function and the proportional adjustment matrix.

5. The block sparse adaptive filter for non-Gaussian noise environment according to claim 4, characterized in that The determining of a filter vector update formula based on the cost function and the proportional adjustment matrix includes: The cost function is minimized by using stochastic gradient descent, and a filter vector update formula is determined in combination with the proportional adjustment matrix.

6. The block sparse adaptive filter for non-Gaussian noise environment according to claim 1, characterized in that The iteration termination requirement refers to that the number of iterations reaches a maximum number of iterations or the difference is less than a set difference threshold.

7. A method for optimizing a block sparse adaptive filter in a non-Gaussian noise environment, applicable to the block sparse adaptive filter in a non-Gaussian noise environment according to any one of claims 1 to 6, characterized in that: include: Obtaining a mixed norm based on the filter vector after uniform block division; Setting an optimal value based on the input vector of the filter and the filter vector to obtain a target output vector; Determining a filter vector update formula based on an adaptive filtering algorithm of arctangent minimum mean square error, wherein the filter vector update formula includes a proportional adjustment matrix, and the proportional adjustment matrix is calculated by the mixed norm; Obtaining a current output vector based on an input vector of the filter and a current filter vector; If the difference between the current output vector and the target output vector does not meet the iteration termination requirement, the current filter vector is updated using the filter vector update formula until the iteration termination requirement is met, and the actual optimal value of the filter vector is output to update the filter vector, thereby obtaining the optimal block sparse adaptive filter.

8. The block sparse adaptive filter optimization method for non-Gaussian noise environment according to claim 7, characterized in that: The method of obtaining a mixed norm based on the filter vector after uniform blocking includes: The 1-norm is calculated for each vector block in the filter vector after uniform blocking, and then the 0-norm is calculated, thereby obtaining the mixed norm.

9. The block sparse adaptive filter optimization method for non-Gaussian noise environment according to claim 7, characterized in that: The adaptive filtering algorithm based on inverse tangent minimum mean square error determines the filter vector update formula, including: A cost function is obtained based on an adaptive filtering algorithm of arctangent minimum mean square error; and a filter vector update formula is determined based on the cost function and the proportional adjustment matrix.

10. The block sparse adaptive filter optimization method for non-Gaussian noise environment according to claim 9, characterized in that: The determining of a filter vector update formula based on the cost function and the proportional adjustment matrix includes: The cost function is minimized by using stochastic gradient descent, and a filter vector update formula is determined in combination with the proportional adjustment matrix.

Images