Design method of ultrasonic cross-metal communication channel echo cancellation filter

Abstract

The invention relates to the technical field of ultrasonic communication, in particular to a design method of an ultrasonic cross-metal communication channel echo cancellation filter, which comprises the following steps of: acquiring a composite signal which is received by a receiving end and comprises a direct signal and an echo signal; creating a corresponding filter mask according to the length of a pre-selected filter; generating a Toeplitz matrix of the composite signal, and according to the filter mask, rejecting a column corresponding to a filter coefficient invalid for echo elimination in the Toeplitz matrix to form a new Toeplitz matrix; constructing a sparse optimization problem of a filter coefficient of the filter; solving the sparse optimization problem to obtain a sparse vector, and recovering the sparse vector into a corresponding filter coefficient according to the filter mask; and the filter is designed according to the filter coefficient obtained through recovery, so that the echo signal in the composite signal is filtered through the designed filter, interference can be avoided, and the filtering performance is improved.

Description

Technical Field

[0001] This invention relates to the field of ultrasonic communication technology, and in particular to a design method, system, device, and medium for an echo cancellation filter for ultrasonic cross-metal communication channels. Background Technology

[0002] In industrial applications, it is often necessary to transmit digital signals over metal obstacles. For metal obstacles such as pressure vessels and pipelines where drilling is not permitted, ultrasonic waves are typically the only option for transmitting digital signals. When ultrasonic waves propagate through metal, the receiver receives multiple echoes in addition to the direct signal. The interval between each echo varies depending on the thickness of the metal wall. When transmitting data through a metal wall, the multiple echoes received by the receiver cause intersymbol interference (ISI) and frequency-selective fading.

[0003] There are two common methods for eliminating echoes received by the receiver during ultrasonic cross-metal communication. The first is to transmit a signal at the transmitting end to cancel the echo, which is generally called "hard cancellation". The second is to add an echo-canceling filter at the receiving end, which is generally called "soft cancellation".

[0004] The "hard cancellation" scheme, which cancels echoes at the transmitting end, requires channel condition information to generate the signal that cancels the echo. In actual communication, the receiving end needs to achieve bidirectional communication to obtain the channel condition information, and tests have shown that its effectiveness is not ideal in some practical test scenarios. Specifically, the premise and theoretical basis of this scheme is that the channel characteristics from the direct signal received by the receiving end to the first received echo signal (first echo signal) are the same as those from the first echo signal to the second echo signal. This channel characteristic can be used to identify the signal used to cancel the echo. However, in actual test scenarios, this premise is not always met, so the "hard cancellation" scheme cannot be used in some environments.

[0005] The "soft cancellation" scheme filters out echoes at the receiving end. The typical approach is to use the received signal as the input signal and the received signal with the echo set to 0 as the target signal, then perform adaptive filtering. Least mean square error adaptive filtering is usually chosen. However, this scheme learns certain noise characteristics, which may lead to additional interference during actual communication when the transmitted and received signals differ from the training signal. Summary of the Invention

[0006] The purpose of this invention is to provide a design method, system, device, and medium for echo cancellation filters in ultrasonic cross-metal communication channels, which can solve the problem that the learning noise characteristics may be interfered with when filtering echoes in "soft cancellation" schemes, thereby affecting the filtering performance.

[0007] To address the aforementioned technical problems, embodiments of the present invention provide a design method for an echo cancellation filter in an ultrasonic cross-metal communication channel, comprising the following steps: Acquire the composite signal received by the receiving end, which includes both direct and echo signals; Based on the pre-selected filter length, a corresponding filter mask is created. The filter mask is used to indicate the position information of the filter coefficients that are effective and ineffective in eliminating echoes. Generate the Toeplitz matrix of the composite signal, and according to the filter mask, remove the columns corresponding to the filter coefficients that are ineffective in eliminating echoes from the Toeplitz matrix to form a new Toeplitz matrix; The sparse optimization problem of filter coefficients is constructed. The sparse optimization problem consists of a least squares term and a sparse regularization term. The least squares term is used to measure the energy difference between the actual direct signal and the ideal direct signal received by the receiver after the filter is filtered using the current filter coefficients. The sparse regularization term is used to sparsify all filter coefficients corresponding to the new Toeplitz matrix and balance the sparsity of the filter coefficients with the filtering degree of the filter based on the current filter coefficients. The sparse optimization problem is solved to obtain sparse vectors, and the sparse vectors are restored to the corresponding filter coefficients according to the filter mask. The filter is designed based on the recovered filter coefficients to filter the echo signal in the composite signal.

[0008] Furthermore, the sparse optimization problem is as follows: ; In the formula, w is the optimization variable, when w = hour, The value is the smallest. Least squares term, Let X be the sparse regularization term, X be the new Toeplitz matrix, and y be the ideal direct signal. , which is the total multiple of the regularization coefficient. The product of λ and λ is the regularization coefficient, 0 < λ / 2. ≤1 is the exponential parameter, d is the length of variable w, and w of length d contains d elements.

[0009] Furthermore, solving the sparse optimization problem includes: In each iteration, solve the following linear equation: ; In the formula, v i (k) =||w i(k-1) |+ε| p-2 k is the number of iterations, ε is a preset value, and β (k) = a i + b i (k) , a i b is the regularization coefficient multiplier determined based on the position of the filter coefficients. i Let diag(v) be the penalty coefficient based on the domain rule. (k) ⊙ β (k) ) indicates that the diagonal element is v (k) ⊙ β (k) A diagonal matrix.

[0010] Furthermore, the size of the Toeplitz matrix is ,in, It is a composite signal. The number of elements in x represents the number of elements in x, and N represents the length of the pre-selected filter.

[0011] Furthermore, the positions in the filter mask that are set to 0 are used to indicate the position information of the filter coefficients that are ineffective in eliminating echoes; The filtering of the echo signal in the composite signal using the designed filter includes: Using sparse vectors as initial values, the least mean square algorithm is used to update the filter coefficients at non-zero positions in the filter mask corresponding to the designed filter.

[0012] Furthermore, the filter coefficients at non-zero positions in the filter mask corresponding to the updated filter design include: If the transducer at the receiving end moves, update the filter coefficients at points 1 to 2 on the left and right of the filter mask corresponding to the designed filter. Otherwise, update the filter coefficients at the points where the filter mask corresponding to the updated filter is set to 1.

[0013] Furthermore, the signal transmitted by the transmitting end corresponding to the receiving end is a linear frequency modulated signal, and the frequency of the linear frequency modulated signal covers the signal bandwidth used by the channel during communication.

[0014] Embodiments of the present invention also provide a design system for an echo cancellation filter in an ultrasonic cross-metal communication channel, comprising: The signal acquisition module is used to acquire the composite signal containing direct signal and echo signal received by the receiver. The mask generation module is used to create a corresponding filter mask based on the pre-selected filter length. The filter mask is used to indicate the position information of the filter coefficients that are effective and ineffective in eliminating echoes. The coefficient initial screening module is used to generate the Toeplitz matrix of the composite signal and, based on the filter mask, remove the columns corresponding to the filter coefficients that are ineffective in eliminating echoes in the Toeplitz matrix to form a new Toeplitz matrix. The coefficient optimization module is used to construct the sparse optimization problem of the filter coefficients. The sparse optimization problem consists of a least squares term and a sparse regularization term. The least squares term is used to measure the energy difference between the actual direct signal received by the receiver and the ideal direct signal after the filter is filtered using the current filter coefficients. The sparse regularization term is used to sparsify all filter coefficients corresponding to the new Toeplitz matrix and balance the sparsity of the filter coefficients with the filtering degree of the filter based on the current filter coefficients. The filter design module is used to solve sparse optimization problems, obtain sparse vectors, and restore the sparse vectors into the corresponding filter coefficients according to the filter mask. The filter is designed based on the recovered filter coefficients to filter the echo signal in the composite signal.

[0015] Embodiments of the present invention also provide a computer device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the above-described ultrasonic cross-metal communication channel echo cancellation filter design method.

[0016] Embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described design method for an echo cancellation filter for an ultrasonic cross-metal communication channel.

[0017] The ultrasonic cross-metal communication channel echo cancellation filter design method provided by this invention has at least the following beneficial effects: First, a filter mask is created based on the filter length. The filter mask indicates the position information of the filter coefficients that are effective and ineffective in eliminating echoes. Based on this, the columns corresponding to the filter coefficients that are ineffective in eliminating echoes in the Toeplitz matrix corresponding to the signal received by the receiver (including the composite signal of direct signal and echo signal) are removed. This reduces the amount of subsequent calculation and eliminates some ineffective filter coefficients in advance. Then, a sparse optimization problem of the filter coefficients is constructed. The sparse optimization problem consists of a least squares term and a sparse regularization term. The least squares term is used to measure the energy difference between the actual direct signal and the ideal direct signal received by the receiver after the filter is filtered with the current filter coefficients. The sparse regularization term is used to sparsify all filter coefficients corresponding to the new Toeplitz matrix and balance the sparsity of the filter coefficients with the filtering degree of the filter based on the current filter coefficients. That is, this sparse optimization problem can achieve further sparse optimization of the filter coefficients. The filter coefficients corresponding to the sparse vector obtained by this sparse optimization problem can effectively eliminate the echo signal received by the receiver.

[0018] Specifically, because there is often some noise or interference in the received signal, the filter coefficients fitted by the least squares term will try to filter out small noise or interference, resulting in additional filter coefficients. The purpose of designing the filter is to filter out echoes, and the noise and interference are not always consistent with those in the training data. By adding a sparse regularization term, we can prevent the learning of noise characteristics, thereby effectively improving the filtering performance. Attached Figure Description

[0019] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0020] Figure 1 A flowchart illustrating the design method of an echo cancellation filter for an ultrasonic cross-metal communication channel provided by the present invention; Figure 2 A schematic diagram of an ultrasonic channel provided by the present invention; Figure 3 A schematic diagram of a signal x received by a receiver according to the present invention; Figure 4 A schematic diagram of a target signal y provided by the present invention; Figure 5 A schematic diagram of filter coefficients and their filtering effect provided by the present invention. Figure 1 ; Figure 6 A schematic diagram of filter coefficients and their filtering effect provided by the present invention. Figure 2 ; Figure 7 A schematic diagram of filter coefficients and their filtering effect provided by the present invention. Figure 3 ; Figure 8 A schematic diagram of filter coefficients and their filtering effect provided by the present invention. Figure 4 ; Figure 9 This is a schematic diagram of the received signal and filtering effect provided by the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0022] The purpose of this invention is to address the problem that ultrasonic waves generate multiple echoes on the metal surface when transmitting digital signals across metal obstacles, causing severe multipath interference to the receiving end. This invention provides a signal processing flow to achieve a highly efficient filter design method for eliminating ultrasonic echoes in cross-metal communication channels. This method can ensure echo elimination performance while maintaining a small computational load and preventing learning noise characteristics.

[0023] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] One embodiment of the present invention relates to a design method for an echo cancellation filter in an ultrasonic cross-metal communication channel. The specific process of the design method for the ultrasonic cross-metal communication channel echo cancellation filter in this embodiment can be described as follows: Figure 1 As shown, it includes: Step 101: Obtain the composite signal received by the receiving end, which includes the direct signal and the echo signal.

[0025] Step 102: Create a corresponding filter mask based on the pre-selected filter length. The filter mask is used to indicate the position information of the filter coefficients that are effective and ineffective in eliminating echoes.

[0026] Step 103: Generate the Toeplitz matrix of the composite signal, and according to the filter mask, remove the columns corresponding to the filter coefficients that are ineffective in eliminating echoes in the Toeplitz matrix to form a new Toeplitz matrix.

[0027] Step 104: Construct the sparse optimization problem of the filter coefficients. The sparse optimization problem consists of a least squares term and a sparse regularization term. The least squares term is used to measure the energy difference between the actual direct signal received by the receiver and the ideal direct signal after the filter is filtered using the current filter coefficients. The sparse regularization term is used to sparsify all filter coefficients corresponding to the new Toeplitz matrix and balance the sparsity of the filter coefficients with the filtering degree of the filter based on the current filter coefficients.

[0028] Step 105: Solve the sparse optimization problem to obtain sparse vectors, and restore the sparse vectors to the corresponding filter coefficients according to the filter mask.

[0029] Step 106: Design a filter based on the recovered filter coefficients to filter the echo signal in the composite signal.

[0030] The following is a detailed description of the implementation details of the ultrasonic cross-metal communication channel echo cancellation filter design method in this embodiment. The following content is only for the convenience of understanding and is not necessary for implementing this solution.

[0031] In step 101, the transmitting end transmits a continuous signal via a piezoelectric transducer. The receiver records the total received signal, including both the direct signal and the echo signal. The signal (i.e., the composite signal) is denoted as signal. The objective of this embodiment is to find In the direct signal section, signals other than the direct signal are set to 0 and denoted as signals. .

[0032] The signal transmitted by the transmitting end is a linear frequency modulated (LFM) signal. The frequency of the LFM signal covers the signal bandwidth used in the channel for communication. When communication uses only a single frequency, such as 2ASK communication, the transmitted signal is a sine wave with a frequency equal to the frequency of the signal used in the communication, ensuring that the filter can properly filter out signals of the frequency used in the communication.

[0033] In step 102, the filter length is set to... Create a filter mask The mask is used to represent the location information of filter coefficients that may exist for echo cancellation, based on prior knowledge of the channel, while reducing the computational load of subsequent steps. For example, if the echo signal is approximately 762 sampling points away from the direct signal, and the echo lasts for 200 sampling points, then when using the direct signal to predict and cancel the echo signal, the current echo signal can be canceled based on the signal approximately 762 points ago, while the signals 300 points ago and 1000 points ago are actually meaningless for canceling the current echo signal.

[0034] Understandably, it is also possible to create a mask that is all 1s, i.e. all of them are marked as valid, and the subsequent algorithm will automatically summarize the positions of the valid filter coefficients.

[0035] In step 103, a signal is generated. The Toeplitz matrix, denoted as ,for For each column, only retain The columns that are non-zero are recombined into a new matrix, denoted as . This forces the ignoring of filter coefficients that, as confirmed by prior knowledge, cannot be used to eliminate echoes, thereby reducing subsequent computational load and preemptively eliminating some invalid coefficients. The Toeplitz matrix is ​​one such example. The size is The size of X is 0, where, It is a composite signal. This represents the number of elements in x. 0 represents the number of points with a value of 1 in the mask m.

[0036] In steps 104 and 105, in order to obtain the filter coefficients that have the greatest impact on echo cancellation and their locations, i.e., to eliminate more invalid coefficients, the following optimization problem (i.e., sparse optimization problem) is solved: ; In the formula, Let w be the sparse vector obtained after optimization, and w be the optimization variable. When w = hour, The value is the smallest. For least squares terms, Let X be the sparse regularization term, X be the new Toeplitz matrix, and y be the ideal direct signal. For exponential parameters, 0 < 0 should be satisfied. ≤1, so that the obtained coefficient It is sparse. , which is the total multiple of the regularization coefficient. The product of λ and λ is the regularization coefficient, calculated based on prior knowledge and the previous iteration. We obtain that d is the length of variable w, and w of length d contains d elements.

[0037] The objective function is This optimization problem adds a penalty term to the sum of the p-th powers of the filter coefficients, where p is between 0 and 1. While ensuring that the filter can effectively filter out echoes, it forces most coefficients to approach 0, making the coefficients that have a smaller effect on eliminating echoes 0.

[0038] To make the objective function smaller, This will satisfy the condition that the sum of the squares of all terms in the residual of the filtering result is equal to the sum of the squares of all terms. (The first term of the objective function) and the absolute values ​​of each term of w itself Weighted sum of powers When the sum of the two terms (the second term of the objective function) is minimized, the coefficients that have little or no impact on filtering, as well as the coefficients whose effects cancel each other out, are all reduced to 0. Therefore, the second term of the objective function decreases, but the first term remains essentially unchanged and is automatically calculated to be 0. Additionally, in the input... There is often some noise or interference in the data. If this penalty term is not added, the fitted filter coefficients will also try to filter out small noise or interference, resulting in additional filter coefficients. The purpose of designing this filter is to filter out echoes, and the noise and interference are not always consistent with those in the training data. If no penalty term is added, the actual filtering effect will be poor. Adding a penalty term can also prevent the learning of noise characteristics. In actual tests, it was also found that in environments with poor signal-to-noise ratio, adding this penalty term can effectively improve the signal-to-noise ratio.

[0039] When solving sparse optimization problems, The initial value for the first iteration is used in the iterative calculation. The least squares solution, the k-th iteration is regarded as Where k is the number of iterations, and ε is a preset value to prevent a fixed value divided by zero; it can be 0.00001. a i The regularization coefficient ratio is determined based on the position of the filter coefficients. The penalty coefficient is based on neighborhood rules: if or Then take b is a pre-defined weight for penalizing adjacent coefficients that are greater than 0; otherwise, it is taken as a negative value. By increasing the penalty factor for other coefficients near the used coefficient, the filter coefficients are forced to be sparser. The reason for using this method is that the ultrasonic echo channel is sparse, and there should be no other coefficients near each coefficient.

[0040] That is, each iteration solves the following linear equation until the parameter change is small: ; Its principle is to make Minimum, then + The minimum, if we use the value from the previous iteration. calculate In this case, the function takes the form of ridge regression, and the result can be calculated directly. The corresponding function that needs to be solved is the linear equation described above.

[0041] Finally, According to the mask Restored to a length of filter coefficient vector This vector is a practically interposable signal. The filter coefficients for eliminating echoes after convolution.

[0042] In step 106, using the sparse vector as the initial value, the Least Mean Squares (LMS) algorithm is employed to update the filter coefficients at the non-zero positions in the filter mask corresponding to the designed filter. Based on... The non-zero values ​​and their positions are used to obtain a new mask. , To indicate The vector of the non-zero point position, i.e. That is, based on the filter coefficients used at this stage, the filter structure to be used in subsequent filtering is determined. As a reference for filter structure, use This serves as the initial value for the filter. When the channel is used for communication, during the channel training phase, the target signal is set to the received signal that retains only the direct signal, and then updated. The sparse filter coefficients at non-zero locations are filtered using the least mean square algorithm to ensure that the filter can adapt to channel changes.

[0043] In the specific implementation, if the transducer at the receiving end moves, the filter coefficients at points 1 to 2 on either side of the filter mask corresponding to the designed filter are updated; otherwise, the filter coefficients at points 1 in the filter mask corresponding to the designed filter are updated. The filter coefficients only include... The coefficients at the value of 1 are updated only at the corresponding positions during filter coefficient updates. When channel conditions may change significantly, such as the possibility of transducer relocation, the updated coefficient range includes... The coefficients at the value of 1, and at points 1 to 2 to the left and right, are updated within the range of values ​​including... The purpose of setting the coefficient to around 1 is to prevent the original filter structure from failing to achieve filtering properly when the transducer is displaced.

[0044] It should be noted that the above filter results can be applied to eliminate signals of arbitrary shapes transmitted by the transmitter, and the coefficient results remain valid for a relatively long period of time. Specifically, when the coupling agent is epoxy resin, the transmitter transducer and the receiver transducer cannot be displaced relative to each other, and the channel conditions remain basically unchanged. It has been tested that the filtering effect remains basically unchanged after 24 hours of retesting after the filter coefficients are measured. If the transducer is reinstalled, all the above steps need to be repeated.

[0045] Figure 2 This is a schematic diagram of a common ultrasonic channel, in which... This represents the channel response of the transducer at the transmitting end, which converts electrical signals into acoustic signals. This represents the channel response of ultrasonic waves propagating in metal. The channel response representing the echo. This represents the channel response of the transducer at the receiving end, which converts acoustic signals into electrical signals. The channel transfer function can then be expressed as: To eliminate the echo response of the channel, a channel response is added at either the transmitting or receiving end. If the filter is applied, the overall channel response is equivalent to eliminating the echo-induced signal. The impact. It is known that the transmitted signal spectrum is... At that time, the direct signal can be represented as The first echo can be represented as That is, the channel response from the direct signal to the first echo can be expressed as: It is exactly the required filter. An important component. Similarly, it can be deduced that the channel response from the nth echo to the (n+1)th echo can also be expressed as... In the above derivation, the obtained filter is not affected by the spectrum of the transmitted signal itself. The training signal is theoretically unaffected by the transmitted signal and can achieve the function with any signal. It can be applied to any transmitted signal. To ensure effectiveness, if the transmitting end may transmit signals of multiple frequencies in actual testing, a linear frequency modulated signal is sent during training to cover the frequency of the transmitted signal.

[0046] Compared with existing technologies, this invention guarantees the performance of echo cancellation, avoids the problem of learning noise characteristics in existing solutions, reduces the resource consumption of echo cancellation filters used in ultrasonic cross-metal communication channels, and retains its ability to adapt to channel changes.

[0047] The following specific embodiment illustrates the design method of the ultrasonic cross-metal communication channel echo cancellation filter of the present invention: First, the ultrasonic cross-metal communication channel described in this invention includes, but is not limited to, aluminum plates, steel plates, etc., which can be used as channels in ultrasonic cross-metal communication scenarios.

[0048] A system for transmitting data via ultrasound through metal uses a 5MHz transducer that communicates through a 40mm thick aluminum plate. The transducer is coupled to the aluminum plate via epoxy resin. The system's transmitter can transmit 2ASK modulated signals or other custom signals, while the receiver can acquire received signals at a 120MHz sampling rate.

[0049] Considering the conditions of the ultrasonic channel, such as Figure 2 As shown, the expression for its channel can be represented as: ; Therefore, a channel response can be designed as follows: The filter is used to eliminate the echo. Actual tests show that using only an FIR filter whose length can cover one echo is sometimes ineffective. For example, in actual testing, it was found that the transducer coupling was not good. In this case, after using the filter of this length, a small amount of second echo signal will remain even though the first echo signal is filtered out well. Therefore, in this embodiment, the FIR filter covers at least three echo lengths.

[0050] Step 1: Considering the aluminum plate thickness is 40mm, the speed of sound in aluminum is approximately 6300m / s, and the receiver sampling rate is 120MHz. The number of sampling points between the echo and the previous waveform is approximately 40mm / (6300m / s)*1 / (120MHz)≈762 points. The goal is to collect the direct signal and the four most recent echo waveforms; therefore, the receiver can collect 4000 signal points.

[0051] Step 2: The transmitting end transmits a 5MHz sine wave signal lasting 2μs, and the receiving end records the received signal as... Collect the most recent 4000 points, such as Figure 3 As shown.

[0052] Step 3: Set the receiver filter length to 3000 points to cover at least 3 to 4 echoes.

[0053] Step 4: Create The Toeplitz matrix, denoted as .in, For one A matrix whose i-th column consists of i zeros and 0s from top to bottom. The first 3000-i points, where i is from 1 to 3000.

[0054] Step 5: The direct signal portion is set to 0, resulting in ,like Figure 4 As shown.

[0055] Step 6: Considering that the number of sampling points between the echo and the previous waveform is approximately 762, and assuming that the filter has 300 points in the corresponding range, let the first term of m be 1. The effective filter for other parts is to set the surrounding 300 points to 1 every 762 points, and the other points to 0. Specifically, in this embodiment, let m = [1,zeros(1,611), ones(1,300), zeros(1,612), ones(1,300), zeros(1,612), ones(1,300), zeros(1,264)], where zeros(1,n) and ones(1,n) represent n consecutive 0s or 1s respectively.

[0056] Step 7: [Regarding...] Only the columns where m is 1 are retained, resulting in a total of 901 columns. At this time, X is a The matrix.

[0057] Step 8: To minimize resource usage, the penalty multiplier for areas with significant time delays is increased. Considering channel conditions and the masking, the penalty coefficient vector is set... [0,zeros(1,611), ones(1,300), zeros(1,612), ones(1,300)*20, zeros(1,612), ones(1,300)*20, zeros(1,264)], keeping only the columns where m is 1, to obtain the penalty coefficient vector in subsequent algorithms. .

[0058] Step 9: Calculation Least square solution , .

[0059] Step 10: Iterative Calculation ,in .in The penalty coefficient is based on neighborhood rules: if or Then take Otherwise take The purpose of this penalty is to prevent filter coefficients from appearing at multiple consecutive points, while the echo channel response should be sparse. Each iteration solves the following linear equation: until Where p is 0.5, the result Recorded as .

[0060] Step 11: According to the mask Restored to a length of filter coefficient vector ,in accordance with Create a new mask at non-zero positions. This mask serves as a reference for subsequent filter structures.

[0061] Step 12: Referring to the LMS algorithm, according to... The iterative method of iteration, where , Compared to the common LMS algorithm, here Update only The coefficient at position 1, and The coefficient at the zero point is set to 0, so it does not need to participate in the multiplication and addition operation.

[0062] The steps described in this process can be implemented online in the communication system to obtain the final coefficients. The coefficients and their filtering effect are as follows: Figure 5 As shown, where, There are 10 non-zero coefficients.

[0063] Compared to the LS algorithm, which uses the same number of multipliers but only 10 coefficients, the filter coefficients and their effects are as follows: Figure 6 As shown, it cannot achieve effective filtering.

[0064] Compared to the filter coefficients and filtering effect obtained using the LS algorithm with 3000 coefficients, such as Figure 7 As shown, 3000 filter coefficients are required. It was found that most of the coefficients are unreasonable because the received signal corresponding to the time delay should not affect the current received signal. For example, the current echo can be eliminated by the direct signal arriving at about 762 points before the echo, and the signal at about 400 points before the echo should be unrelated to the current echo. However, the filter coefficients calculated here still have large values, indicating that they have learned the noise characteristics. This means that when there is a lot of noise in the channel, the filtering effect may decrease after a small change in the channel or a change in the transmitted signal.

[0065] When the received signal-to-noise ratio is low, the ability of this invention to prevent learning noise is clearly demonstrated. In this embodiment, the noise in the channel received by the receiver is relatively small compared to many other studies. To simulate a noisier channel, additive white noise with a power of 20dB is added to the received signal after quantization. This is done to filter out noise present in the signal itself. Set all coefficients to 1, calculate the filter coefficients obtained by this invention and the filter coefficients obtained by the LS algorithm with 3000 coefficients, and compare them to the filtered signal obtained by applying additive white noise to the received signal of the same signal. Figure 8 As shown, the filter coefficients obtained by the present invention have a total of 83 non-zero coefficients. The signal after the direct signal is regarded as noise. The noise of the signal obtained by the present invention is 33.39% of that of the traditional scheme, that is, the signal-to-noise ratio is improved by 4.76dB.

[0066] Even if the transmitted signal is different from the training signal, the filter obtained by this invention can still achieve a good filtering effect. Figure 9 The filtering effect was demonstrated when the transmitted signal used during training differed from the transmitted signal at the transmitter during actual operation, indicating that the previously described method is indeed unaffected by the transmitted signal.

[0067] The steps of the various methods described above are only for clarity. In practice, they can be combined into one step or some steps can be split into multiple steps. As long as they include the same logical relationship, they are all within the protection scope of this invention. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, without changing the core design of the algorithm and process, are also within the protection scope of this invention.

[0068] Another embodiment of the present invention relates to a design system for an echo cancellation filter in an ultrasonic cross-metal communication channel. The implementation details of this ultrasonic cross-metal communication channel echo cancellation filter design system are described below. The following details are provided for ease of understanding and are not essential for implementing this solution. The ultrasonic cross-metal communication channel echo cancellation filter design system of this embodiment includes: The signal acquisition module is used to acquire the composite signal containing direct signal and echo signal received by the receiver. The mask generation module is used to create a corresponding filter mask based on the pre-selected filter length. The filter mask is used to indicate the position information of the filter coefficients that are effective and ineffective in eliminating echoes. The coefficient initial screening module is used to generate the Toeplitz matrix of the composite signal and, based on the filter mask, remove the columns corresponding to the filter coefficients that are ineffective in eliminating echoes in the Toeplitz matrix to form a new Toeplitz matrix. The coefficient optimization module is used to construct the sparse optimization problem of the filter coefficients. The sparse optimization problem consists of a least squares term and a sparse regularization term. The least squares term is used to measure the energy difference between the actual direct signal received by the receiver and the ideal direct signal after the filter is filtered using the current filter coefficients. The sparse regularization term is used to sparsify all filter coefficients corresponding to the new Toeplitz matrix and balance the sparsity of the filter coefficients with the filtering degree of the filter based on the current filter coefficients. The filter design module is used to solve sparse optimization problems, obtain sparse vectors, and restore the sparse vectors into the corresponding filter coefficients according to the filter mask. The filter is designed based on the recovered filter coefficients to filter the echo signal in the composite signal.

[0069] It is not difficult to see that this embodiment is a system embodiment corresponding to the above method embodiments, and this embodiment can be implemented in conjunction with the above method embodiments. The relevant technical details and technical effects mentioned in the above embodiments are still valid in this embodiment, and will not be repeated here to reduce repetition. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the above embodiments.

[0070] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by this invention; however, this does not mean that other units are absent from this embodiment.

[0071] Another embodiment of the present invention relates to a computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the ultrasonic cross-metal communication channel echo cancellation filter design method of the above embodiments.

[0072] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0073] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0074] Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the method embodiments described above.

[0075] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0076] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing the present invention, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of the present invention.

Claims

1. An ultrasonic cross-metal communication channel echo cancellation filter design method, characterized by, The method comprises: acquiring a composite signal received by a receiving end, the composite signal comprising a direct signal and an echo signal; creating a filter mask corresponding to a length of a filter selected in advance, the filter mask being used to indicate position information of filter coefficients of the filter which are effective and ineffective in eliminating the echo respectively; generating a Toeplitz matrix of the composite signal, and removing columns corresponding to filter coefficients ineffective in eliminating the echo in the Toeplitz matrix according to the filter mask, to form a new Toeplitz matrix; constructing a sparse optimization problem of filter coefficients of the filter, the sparse optimization problem being composed of a least square term and a sparse regularization term, the least square term being used to measure energy of a difference between an actual direct signal received by the receiving end after the filter with current filter coefficients is used for filtering and an ideal direct signal, and the sparse regularization term being used to sparsify all filter coefficients corresponding to the new Toeplitz matrix and balance a sparsity degree of the filter coefficients and a filtering degree of the filter based on the current filter coefficients; solving the sparse optimization problem to obtain a sparse vector, and restoring the sparse vector into corresponding filter coefficients according to the filter mask; designing the filter according to the restored filter coefficients, to filter the echo signal in the composite signal through the designed filter.

2. The ultrasonic cross-metal communication channel echo cancellation filter design method of claim 1, wherein, The sparse optimization problem is: ； In the formula, w is the optimization variable, when w = hour, The value is the smallest. For least squares terms, Let X be the sparse regularization term, X be the new Toeplitz matrix, and y be the ideal direct signal. , which is the total multiple of the regularization coefficient. The product of λ and λ is the regularization coefficient, 0 < λ / 2. ≤1 is the exponential parameter, d is the length of variable w, and w of length d contains d elements.

3. The ultrasonic cross-metal communication channel echo cancellation filter design method of claim 2, wherein, The solving of the sparse optimization problem comprises: in each iteration, solving the following linear equation: ； In the formula, v i (k) =||w i (k-1) |+ε| p-2 , k is the iteration number, ε is a preset value, β (k) = a i + b i (k) , b i is a penalty coefficient based on domain rules, a i is a regularization coefficient multiple determined according to the position of the filter coefficient, b i is a penalty coefficient based on domain rules, diag(v (k) ⊙ β (k) ) represents a diagonal matrix with v (k) ⊙ β (k) as diagonal elements.

4. The ultrasonic cross-metal communication channel echo cancellation filter design method of claim 1, wherein, The size of the Toeplitz matrix is wherein, is a composite signal, represents the number of elements of x, and N represents the length of the filter selected in advance.

5. The ultrasonic cross-metal communication channel echo cancellation filter design method of claim 1, wherein, The positions set to 0 in the filter mask are used to indicate position information of filter coefficients of the filter which are ineffective in eliminating the echo; The filtering of the echo signal in the composite signal through the designed filter comprises: taking the sparse vector as an initial value, and updating filter coefficients of non-zero positions in the filter mask corresponding to the designed filter by using a least mean square algorithm.

6. The ultrasonic cross-metal communication channel echo cancellation filter design method of claim 5, wherein, The updating of the filter coefficients of the non-zero positions in the filter mask corresponding to the designed filter comprises: if a transducer of the receiving end moves, updating filter coefficients of positions set to 1 and positions of 1 to 2 points left and right in the filter mask corresponding to the designed filter; otherwise, updating filter coefficients of positions set to 1 in the filter mask corresponding to the designed filter.

7. The ultrasonic cross-metal communication channel echo cancellation filter design method of claim 1, wherein, The signal sent by a sending end corresponding to the receiving end is a linear frequency modulation signal, and a frequency coverage of the linear frequency modulation signal covers a signal bandwidth used when a channel is used for communication.

8. An ultrasonic cross-metal communication channel echo cancellation filter design system, characterized by, The system comprises: a signal acquisition module, configured to acquire a composite signal received by a receiving end, the composite signal comprising a direct signal and an echo signal; a mask generation module, configured to create a filter mask corresponding to a length of a filter selected in advance according to the length of the filter, the filter mask being used to indicate position information of filter coefficients of the filter which are effective and ineffective in eliminating the echo respectively; a coefficient preliminary screening module, configured to generate a Toeplitz matrix of the composite signal, and remove columns corresponding to filter coefficients ineffective in eliminating the echo in the Toeplitz matrix according to the filter mask, to form a new Toeplitz matrix; a coefficient optimization module, configured to construct a sparse optimization problem of filter coefficients of the filter, the sparse optimization problem being composed of a least square term and a sparse regularization term, the least square term being used to measure energy of a difference between an actual direct signal received by the receiving end after the filter with current filter coefficients is used for filtering and an ideal direct signal, and the sparse regularization term being used to sparsify all filter coefficients corresponding to the new Toeplitz matrix and balance a sparsity degree of the filter coefficients and a filtering degree of the filter based on the current filter coefficients; and solve the sparse optimization problem to obtain a sparse vector, and restore the sparse vector into corresponding filter coefficients according to the filter mask; and a filter design module, configured to design the filter according to the restored filter coefficients, to filter the echo signal in the composite signal through the designed filter. The coefficient optimization module is configured to construct a sparse optimization problem of filter coefficients of the filter, the sparse optimization problem being composed of a least square term and a sparse regularization term, the least square term being used to measure energy of a difference between the actual direct signal received by the receiving end and the ideal direct signal after the filter performs filtering by using the current filter coefficients, and the sparse regularization term being used to sparsify all filter coefficients corresponding to the new Toeplitz matrix and balance the sparsity of the filter coefficients and the filtering degree of the filter based on the current filter coefficients. The filter design module is configured to solve the sparse optimization problem to obtain a sparse vector, and restore the sparse vector into corresponding filter coefficients according to a filter mask. The filter is designed according to the restored filter coefficients, so that the echo signal in the composite signal is filtered by the designed filter.

9. A computer device, comprising: Comprise: At least one processor; And a memory connected in communication with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the ultrasonic cross-metal communication channel echo cancellation filter design method in any one of claims 1 to 7.

10. A computer readable storage medium storing a computer program, characterized in that, The computer program is executed by the processor to implement the ultrasonic cross-metal communication channel echo cancellation filter design method in any one of claims 1 to 7.

Images