Self-interference removal method under influence of ultrasonic interference

By combining polynomial nonlinear model and NLMS adaptive algorithm, the problem of high-order nonlinear distortion in nonlinear systems is solved, and the removal accuracy of ultrasonic interference signals and recording audio quality are significantly improved.

CN118921597APending Publication Date: 2024-11-08UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410967845.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The existing self-interference removal algorithm cannot effectively capture the high-order nonlinear distortion in ultrasonic interference signals in nonlinear systems, resulting in poor recording audio quality.

Method used

A nonlinear model based on polynomials is used to estimate the higher-order coefficients of the signal through the NLMS adaptive algorithm, and a more accurate nonlinear system model is constructed. After filtering, the NLMS algorithm is used to remove the self-interference signal.

Benefits of technology

The accuracy of self-interference signal removal and recording audio quality are improved, especially at high transmit power, which significantly improves the cancellation effect of interfering signal.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118921597A_ABST
    Figure CN118921597A_ABST
Patent Text Reader

Abstract

The invention relates to a method for removing self-interference under the influence of ultrasonic interference, which belongs to the field of self-interference elimination and comprises the following steps of: calculating a theoretical value Sout (t) of an interference signal received by a microphone based on a polynomial nonlinear model Sout (t) = A1S (t) + A2S2 (t) + A3S3 (t) +... ANSN (t), and estimating a tap vector in a self-adaptive filtering process through an NLMS (Non-Least Mean Square) self-adaptive algorithm; filtering the obtained theoretical value Sout (t) of the interference signal to obtain an interference signal x (n) in the recording signal; and with the filtered interference signal estimation value x (n) as a reference signal, performing self-interference signal elimination by using an NLMS adaptive algorithm. According to the method, nonlinear characteristics generated in a transmitting link are analyzed, a higher-order nonlinear distortion item is considered, and nonlinear behaviors in the ultrasonic jammer can be simulated and estimated more accurately, so that the accuracy of self-interference signal removal is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of self-interference elimination, and in particular to a method for removing self-interference under the influence of ultrasonic interference. Background Art

[0002] Ideally, the microphone and the speaker are both linear systems, so when there is an input S(t), the system output can be considered to be in the form of S out (t) = A1*S(t), where A1 is the gain of the signal after passing through the amplifier. However, in reality, the acoustic elements (such as amplifiers, etc.) in microphones and speakers are linear only in the audible frequency range (<20kHz). They exhibit nonlinear characteristics in the ultrasonic frequency band (>20kHz). Therefore, for ultrasonic signals, the output signal after passing through the amplifier is in the form of S out (t) = A1S(t) + A2S 2 (t)+A3S 3 (t)+…≈A1S(t)+A2S 2 (t), where A1, A2, A3, ... are the gains of the signals after passing through the amplifier.

[0003] Most of the existing algorithms for self-interference removal using NLMS (normalized least mean square adaptive algorithm) adaptive filters use the transmitted interference signal (i.e., S(t)) as the reference signal. However, in actual situations, the output signal of a nonlinear system will produce additional harmonics and intermodulation distortion terms in addition to S(t), such as the high-order component S in another formula: 2 (t), etc. Therefore, the method of using S(t) as a reference signal is not applicable to nonlinear systems. In the adaptive filtering algorithm for nonlinear systems, existing studies have used the second-order simplified Voltra series to make nonlinear estimates of interference signals (i.e., A1S(t)+A2S 2 (t)) and input it into the adaptive filter as a reference signal. However, this is unreasonable because for ultrasonic interference equipment, if the equipment itself wants to obtain clear recorded audio, it needs a more accurate reference signal to accurately estimate the interference signal generated during the transmission process. Considering only the quadratic term will make the model unable to fully capture the higher-order nonlinear distortion that may be generated during the transmission process. These distortions are particularly significant at high transmission power. Therefore, how to solve this problem is currently under consideration. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art, provide a method for removing self-interference under the influence of ultrasonic interference, and solve the shortcomings of the prior art.

[0005] The purpose of the present invention is achieved by the following technical solution: a method for removing self-interference under the influence of ultrasonic interference, the method for removing self-interference comprising:

[0006] Step 1: Polynomial-based nonlinear model S out (t) = A1S(t) + A2S 2 (t)+A3S 3 (t)+...A N S N (t) Calculate the theoretical value S of the interference signal received by the microphone out (t), where each coefficient A i Represents the gain of the signal after the amplifier, i∈N, and the tap vector in the adaptive filtering process is estimated by the NLMS adaptive algorithm, that is, the coefficients A i ;

[0007] Step 2: Get the theoretical value S of the interference signal out (t) filtering to obtain the interference signal x(n) in the recording signal;

[0008] Step 3: Use the filtered interference signal estimate x(n) as a reference signal and use the NLMS adaptive algorithm to eliminate the self-interference signal.

[0009] The tap vector estimated in the adaptive filtering process by the NLMS adaptive algorithm includes:

[0010] Initialize the tap vector to 0, and the vector length is the preset filter order, and the iterative equation for the unknown tap vector is: Among them, μ is the step size parameter, δ>0, x(n) is the interference signal at the nth moment, ||x(n)|| 2 represents the second-order norm of x(n), e * (n) represents the error signal, which is obtained by subtracting the actual output signal d(n) from the estimated interference signal y'(n);

[0011] After the tap vector is obtained, the estimated interference signal y'(n) at the nth moment is obtained through the formula y'(n)=w(n)*x(n).

[0012] The step three specifically includes:

[0013] Obtain the estimated self-interference signal y'(n) at this time according to the interference signal x(n);

[0014] According to the signal S collected by the microphone mic (n) minus the self-interference signal y'(n) to obtain the pure human voice signal v(n), that is, v(n) = S mic (n)-y'(n).

[0015] The present invention has the following advantages: a method for removing self-interference under the influence of ultrasonic interference, by analyzing the nonlinear characteristics generated in the transmission link, taking into account higher-order nonlinear distortion terms, can more accurately simulate and estimate the nonlinear behavior in the ultrasonic jammer, thereby improving the accuracy of self-interference signal removal. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the process of the present invention;

[0017] Figure 2 This is a comparison chart of the recording quality of the jammer and the interfered device at the same distance from the speaker;

[0018] Figure 3 This is a comparison chart of the jammer recording effects;

[0019] Figure 4 The following is a comparison chart of PESQ between different systems at an interference distance of 50 cm. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of the present application provided below in conjunction with the drawings is not intended to limit the scope of protection of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application. The present invention is further described below in conjunction with the drawings.

[0021] like Figure 1 As shown, the present invention specifically relates to a method for removing self-interference under the influence of ultrasonic interference, which takes into account higher-order nonlinear distortion terms. By constructing a more accurate nonlinear system model and eliminating the nonlinear characteristics of the transmission link, the ultrasonic jammer can more effectively remove the self-interference signal; the difference from other existing self-interference elimination algorithms is that the method of the present invention takes into account the nonlinear distortion generated when the ultrasonic signal passes through the transmission link. That is, when calculating the reference signal, the formula that takes into account higher-order distortion is used to simulate the output signal S when the ultrasonic signal produces nonlinear distortion. out. When multiple frequency signals pass through the filter at the same time, the output signal will contain components of the original signal as well as new frequency components generated by harmonic distortion and intermodulation distortion. For example, consider a signal containing two components of frequency f1 and f2. When they pass through a nonlinear system, in addition to the original frequency components, they will also generate new frequency components such as frequency-multiplied signals 2f1, 2f2, and difference frequency signals |f1±f2|. By considering the high-order components of the signal, the present invention fully simulates the distorted low-frequency signal generated by the mutual influence of the harmonic components when the ultrasonic signal passes through the transmission link.

[0022] Specifically include the following:

[0023] Step 1: Calculation of the theoretical value of the interference signal: The interference signal S sent by the ultrasonic jammer itself is known in (t), in order to consider the impact of higher-order frequency components on the output signal, it is necessary to use a polynomial-based nonlinear model S out (t) = A1S(t) + A2S 2 (t)+A3S 3 (t)+...A N S N (t) Calculate the theoretical value S of the interference signal received by the microphone out (x), where each coefficient A i Represents the gain of the signal after the amplifier, which is related to the actual propagation path and is estimated by the adaptive filter, i∈N, and the tap vector in the adaptive filtering process is estimated by the NLMS adaptive algorithm, that is, the coefficients A i ;

[0024] Step 2: Filtering: The filtering process is consistent with the processing method of the recording signal. The theoretical value S of the interference signal obtained above is filtered with a cutoff frequency of 20kHz. out (t) Filter and retain only the low-frequency part to obtain the interference signal x(n) in the recorded signal;

[0025] In the specific operation process, a low-pass filter is used to filter the theoretical value of the signal. It removes noise from the signal or extracts specific frequency components from the signal. Parameter 1 is a pre-defined filter, with the passband frequency set to 15kHz and the stopband frequency set to 20kHz. Parameter 2 is the signal to be filtered.

[0026] Step 3: Use the filtered interference signal estimate x(n) as a reference signal and use the NLMS adaptive algorithm to eliminate the self-interference signal.

[0027] Furthermore, the tap vector in the adaptive filtering process is estimated by the NLMS adaptive algorithm including:

[0028] First, the tap vector is initialized to 0, and the vector length is the preset filter order. The iterative equation for the unknown tap vector is obtained as follows: Among them, μ is the step size parameter, which determines the speed and amplitude of the filter parameter update, x(n) is the interference signal at the nth moment, ||x(n)|| 2 represents the second-order norm of x(n), e * (n) represents the error signal, which is obtained by subtracting the actual output signal d(n) from the estimated interference signal y'(n);

[0029] After the tap vector is obtained, the estimated interference signal y'(n) at the nth moment is obtained through the formula y'(n)=w(n)*x(n).

[0030] Furthermore, step three specifically includes:

[0031] Obtain the estimated self-interference signal y'(n) at this time according to the interference signal x(n);

[0032] According to the signal S collected by the microphone mic (n) minus the self-interference signal y'(n) to obtain the pure human voice signal v(n), that is, v(n) = S mic (n)-y'(n).

[0033] In addition to receiving the generated self-interference y(n) and human voice v(n), the jammer will also generate an estimated self-interference signal y'(n) based on the interference signal x(n) through an adaptive filtering algorithm within the system. A pure human voice signal can be obtained by subtracting the received microphone signal from the generated estimated interference signal.

[0034] The present invention prepares a designed interference signal x(n) in advance, and this interference signal will be played when the jammer is started. Before the experiment begins, the filter order is set to 1100 in advance. The filter order is the length of the tap vector in the algorithm process. This is an empirical value obtained by continuously testing the values ​​of different filter orders. At the beginning of the experiment, the jammer will generate an estimated self-interference signal y'(n) based on the interference signal x(n) in advance. During the generation process, the interference signal x(n) enters the tap vector in sequence and multiplies it according to the chronological order. The number of points entering the tap vector is determined by the order, and 1100 points will be calculated at the same time. The estimated self-interference signal y'(n) at the current moment is calculated from these 1100 points. The specific formula is as follows:

[0035] y′(n)=w(n)*x(n)

[0036] After completing an estimation, the algorithm needs to update the weights of the tap vector. The update method is The appropriate step size can help maintain the stability of the algorithm. Therefore, the choice of step size needs to strike a balance between fast convergence and algorithm stability. The present invention uses a dynamic step size to automatically adjust the value to achieve a balance between fast convergence and algorithm stability. The value of the step size parameter μ is automatically updated using the following formula:

[0037]

[0038] where e * (n) represents the difference between the actual received signal and the estimated interference signal, that is:

[0039] e * (n) = y(n) + v(n) - y′(n)

[0040] Where ε(n) is a weighted error vector that measures the mismatch of the tap vector, and the iteration method is as follows:

[0041]

[0042] It is set that the μ value is updated once every 100 estimates of y'(n). Too frequent updates will greatly increase the amount of calculation, making it impossible to meet the requirements of real-time adaptive filtering of the jammer. Too few updates will make the interference removal effect worse. 100 times is an experimental experience value. After obtaining the real-time updated μ value, the tap vector can be iterated. When the interference signal x(n) is fully input into the tap vector, the complete estimation of the interference signal y'(n) is completed. At this time, the pure human voice signal can be obtained by subtracting the estimated interference signal from the signal received by the jammer microphone.

[0043] Part of the experimental results and analysis: In this experiment, the evaluation index of the voice quality of the ultrasonic jammer recording signal is PESQ (Perceptual Evaluation of Speech Quality). PESQ is an objective standard for evaluating speech quality established by the International Telecommunication Union (ITU). This invention will use this objective index to evaluate speech quality.

[0044] Anti-interference ability of adaptive filtering in different languages:

[0045] Test results such as Figure 3 As shown in the figure, it is shown that after using the above adaptive filtering algorithm, the system can obtain a pure human voice signal from the self-interfered speech according to the estimated interference signal. By comparing the spectrum of the human voice and the spectrum of the interference signal, it can be seen that the interference signal covers all the information in the human voice frequency band and effectively interferes with the speech signal. From the spectrum of the recorded speech in the device, it can be seen that the influence of the interference signal on the speech signal is basically eliminated in the frequency domain. The filtering of the human voice signal is achieved.

[0046] The relationship between self-interference cancellation capability and speaker distance:

[0047] A one-minute audio clip of the English News Broadcast was played using a loudspeaker simulating the volume of a human voice (65-70 decibels), and the jammer and unauthorized recorder were gradually moved away from the loudspeaker in increments of 0.2 meters. The present invention conducted this test within a 10-meter radius of the loudspeaker. The experimental results are shown in Figure 1. Figure 2 As shown. Within a 2-meter radius of the speaker, there is a significant difference in recording quality between the recording device using the solution of the present invention and the unauthorized recorder, with the maximum PESQ value difference being 1.9.

[0048] However, as the interference distance gradually increases, although the jammer's recording effect is still better than that of the unauthorized device, its voice quality begins to decline significantly. When the distance from the speaker exceeds 8 meters, when the distance between the jammer and the unauthorized recorder reaches 16 meters, the recording quality of the two is almost the same, and the maximum difference in PESQ is only 0.7. This is mainly because the ultrasonic interference signal decays faster than the low-frequency human voice. In the experimental scenario of the present invention, when the interference recording is twice the recording distance, the faster attenuation of the ultrasonic signal will cause the interference effect to drop sharply.

[0049] Comparison of PESQ voice quality between different systems:

[0050] Prior to this, the experiment of anti-interference technology evaluated the PESQ index when the distance between the speaker and the jammer was 50 cm. Therefore, the present invention also uses the PESQ value within 50 cm for comparative analysis. Figure 4 It can be seen that the maximum PESQ of other systems is between 2.04 and 2.12, while the maximum PESQ value of the jammer of the anti-interference algorithm proposed in the present invention can reach 2.7, an increase of 0.6, which is equivalent to an improvement of nearly 30% in the recording effect of the jammer. This is mainly because the algorithm of the present invention takes into account the high-order harmonics generated when the ultrasonic signal passes through the nonlinear system, making the estimation of the interference signal more accurate and the interference removal effect better.

[0051] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used for various other combinations, modifications and improvements, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art do not deviate from the spirit and scope of the present invention, and should be within the scope of protection of the claims attached to the present invention.

Claims

1. A method for removing self-interference under the influence of ultrasonic interference, Features: The self-interference removal method comprises: Step 1: Polynomial-based nonlinear model S out (t) = A 1 S(t)+A 2 S 2 (t)+A 3 S 3 (t)+…A N S N (t) Calculate the theoretical value S of the interference signal received by the microphone out (t), where the coefficients Ai represent the gain of the signal after the amplifier, i∈N, and the tap vector in the adaptive filtering process is estimated by the NLMS adaptive algorithm, that is, the coefficients A i ; Step 2: Get the theoretical value S of the interference signal out (t) filtering to obtain the interference signal x(n) in the recording signal; Step 3: Use the filtered interference signal estimate x(n) as a reference signal and use the NLMS adaptive algorithm to eliminate the self-interference signal.

2. The method for removing self-interference under the influence of ultrasonic interference according to claim 1, Features: The tap vector estimated in the adaptive filtering process by the NLMS adaptive algorithm includes: Initialize the tap vector to 0, and the vector length is the preset filter order, and the iterative equation for the unknown tap vector is: Among them, μ is the step size parameter, δ>0, x(n) is the interference signal at the nth moment, ||x(n)|| 2 represents the second-order norm of x(n), e * (n) represents the error signal, which is obtained by subtracting the actual output signal d(n) from the estimated interference signal y'(n); After the tap vector is obtained, the estimated interference signal y'(n) at the nth moment is obtained through the formula y'(n)=w(n)*x(n).

3. The method for removing self-interference under the influence of ultrasonic interference according to claim 2, Features: The step three specifically includes: Obtain the estimated self-interference signal y'(n) at this time according to the interference signal x(n); According to the signal S collected by the microphone mic (n) minus the self-interference signal y'(n) to obtain the pure human voice signal v(n), that is, v(n) = S mic (n)-y′(n).