A method and device for analyzing anti-interference performance of blind source signal separation communication

Abstract

The application relates to a blind source signal separation communication anti-interference performance boundary analysis method and device. The method comprises the following steps: constructing multiple groups of received signal vectors; for each group of received signal vectors, calculating the partial derivative of a log-likelihood function to a mixing matrix; taking the statistical expectation of all the partial derivatives as a Monte Carlo estimation value of a Fisher information matrix; defining a gain matrix and calculating a Jacobian matrix mapped from the mixing matrix to the gain matrix; using the inverse matrix of the Fisher information matrix and the Jacobian matrix, calculating the Cramer-Rao lower bound of the power ratio of residual interference signals and communication signals, and outputting the lower bound as the performance limit of blind source signal separation communication anti-interference. The method can effectively evaluate the separation performance limit of the blind source signal separation communication anti-interference algorithm under a noise condition.

Description

Technical Field

[0001] This application relates to the field of wireless communication anti-interference technology, and in particular to a method and apparatus for analyzing the anti-interference performance of blind source signal separation communication. Background Technology

[0002] Blind source signal separation refers to a class of theories and techniques that estimate source signals and their mixing processes using only observed mixed signals when there is little or no prior information. In recent years, blind source signal separation has gained widespread attention in signal processing fields such as underwater acoustics, biomedicine, and image processing. Simultaneously, due to the inherent similarities between wireless communication signal separation and speech signal separation, blind source signal separation technology has been rapidly applied to the problem of mixing and separating radio signals, opening up a new technical route for communication anti-interference. Blind source signal separation communication anti-interference methods focus on the statistical domain dimension of the signal, requiring no recognition of the interfering signal and no additional spectrum resources. It can separate independent communication signals from interfering signals at the receiving end, achieving effective extraction of the communication signal and thus improving the communication system's ability to cope with high-power suppressive interference.

[0003] However, although blind source signal separation technology can significantly enhance communication anti-interference performance, its separation performance limit under noisy conditions is still unclear, which is one of the current difficulties in this field and to some extent restricts the design and practical application of blind source signal separation anti-interference algorithms. Summary of the Invention

[0004] Therefore, it is necessary to provide a method and apparatus for analyzing the anti-interference performance limits of blind source signal separation communication, which can solve the technical problem of unclear performance limits under noisy conditions.

[0005] A method for analyzing the anti-interference performance limits of blind source signal separation communication, the method comprising: Multiple sets of received signal vectors are constructed. Each set of received signal vectors is obtained by mixing the source signal vectors through a mixing matrix and then adding noise. For each set of received signal vectors, calculate the partial derivative of the log-likelihood function with respect to the mixing matrix; The statistical expectation of all partial derivatives is used as the Monte Carlo estimate of the Fisher information matrix; Define the gain matrix based on the estimated mixing matrix and the ideal separation matrix, and calculate the Jacobian matrix required to map from the mixing matrix to the gain matrix; Using the inverse of the Fisher information matrix and the Jacobian matrix, the Cramer-Rao lower bound of the ratio of residual interference signal power to communication signal power is calculated, and the Cramer-Rao lower bound is output as the performance limit of blind source signal separation communication anti-interference.

[0006] A device for analyzing the anti-interference performance of blind source signal separation communication, the device comprising: The signal construction module is used to construct multiple sets of received signal vectors. Each set of received signal vectors is obtained by mixing the source signal vectors through a mixing matrix and then adding noise. The partial derivative calculation module is used to calculate the partial derivative of the log-likelihood function with respect to the mixing matrix for each set of received signal vectors. The information matrix estimation module is used to take the statistical expectation of all partial derivatives as the Monte Carlo estimate of the Fisher information matrix; The mapping transformation module is used to define the gain matrix based on the estimated mixing matrix and the ideal separation matrix, and to calculate the Jacobian matrix required to map from the mixing matrix to the gain matrix; The performance limit output module is used to calculate the Cramer-Rao lower bound of the ratio of residual interference signal power to communication signal power using the inverse matrix of the Fisher information matrix and the Jacobian matrix, and outputs the Cramer-Rao lower bound as the performance limit of blind source signal separation communication anti-interference.

[0007] The aforementioned method and apparatus for analyzing the anti-interference performance limits of blind source signal separation communication utilize the Monte Carlo method to calculate the Cramer-Rao lower bound of the mixing matrix in noisy blind source signal separation scenarios. A gain matrix is ​​defined, with the size of its off-diagonal elements measuring the power ratio (ISR) of the residual interference signal to the communication signal after blind source signal separation. The Cramer-Rao lower bound of the mixing matrix is ​​extended to the Cramer-Rao lower bound of the ISR index, thereby analyzing and obtaining the performance limit of blind source signal separation communication anti-interference. This application directly derives the Cramer-Rao lower bound of ISR under a noisy model, while existing methods typically ignore the influence of noise. Therefore, the performance limit analysis results provided by this application are more accurate and closely approximate the performance obtained from actual separation algorithms. This provides a theoretical basis and technical support for the research of blind source signal separation communication anti-interference methods, solving the technical problem of unclear performance limits under noisy conditions. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating a method for analyzing the anti-interference performance of blind source signal separation communication in one embodiment. Figure 2 This is a schematic diagram of the simulation experiment results in one embodiment; Figure 3 This is a structural block diagram of a blind source signal separation communication anti-interference performance boundary analysis device in one embodiment. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0010] A typical application scenario of this application is the boundary analysis of communication anti-interference performance in a scenario comprising a communication signal transmitter, an interference signal transmitter, and a dual-channel receiving terminal. The receiving end utilizes blind source signal separation technology to estimate the separation matrix by receiving the mixed signal, thereby separating the communication signal from the interference signal to achieve communication anti-interference. The main parameters are defined as follows: the source signal composed of the communication signal and the interference signal is denoted as... The mixing matrix is ​​denoted as The received signal is denoted as .

[0011] In one embodiment, such as Figure 1 As shown, a method for analyzing the anti-interference performance boundary of blind source signal separation communication is provided, including the following steps: Step 102: Construct multiple sets of received signal vectors. Each set of received signal vectors is obtained by mixing the source signal vectors with a mixing matrix and then adding noise.

[0012] By constructing multiple sets of received signal vectors, a data foundation is provided for subsequent calculations of partial derivatives and the Fisher information matrix using the Monte Carlo method.

[0013] Step 104: For each set of received signal vectors, calculate the partial derivative of the log-likelihood function with respect to the mixing matrix; use the statistical expectation of all partial derivatives as the Monte Carlo estimate of the Fisher information matrix.

[0014] The partial derivatives are fundamental parameters for subsequent calculations of the Fisher information matrix. The Fisher information matrix contains information about the separation matrix and is a key parameter for calculating the Cramer-Rao bounds in estimating the separation matrix. The Monte Carlo method can effectively estimate this matrix and is suitable for mixed scenarios where the source signal constellation points are discrete and the dimension is low.

[0015] Step 106: Define the gain matrix based on the estimated mixing matrix and the ideal separation matrix, and calculate the Jacobian matrix required to map from the mixing matrix to the gain matrix.

[0016] Gain matrix The size of its off-diagonal elements directly measures the power ratio (ISR) of the residual interference signal to the communication signal after blind source signal separation. The Jacobian matrix reflects the power ratio of the residual interference signal to the communication signal after blind source signal separation. The transformation mapping required to extend the Cramérault lower bound of the ISR to the Cramérault lower bound.

[0017] Step 108: Using the inverse matrix of the Fisher information matrix and the Jacobian matrix, calculate the Cramer-Rao lower bound of the ratio of residual interference signal power to communication signal power, and output the Cramer-Rao lower bound as the performance limit of blind source signal separation communication anti-interference.

[0018] The mixing matrix The lower bound of Cramer-Rao is extended to the lower bound of the ISR index, and then the performance limit of anti-interference in blind source signal separation communication is analyzed, providing theoretical basis and technical support for the research of anti-interference methods in blind source signal separation communication.

[0019] This application constructs multiple sets of received signal vectors and calculates the partial derivative of the log-likelihood function with respect to the mixing matrix based on each set of received signal vectors. Then, it uses the statistical expectation of all partial derivatives as the Monte Carlo estimate of the Fisher information matrix, thus directly obtaining the Cramer-Rao lower bound of the mixing matrix under noisy conditions. Furthermore, it defines a gain matrix to quantify the power ratio (ISR) of the residual interference signal to the communication signal and calculates the Jacobian matrix mapping from the mixing matrix to this gain matrix, thereby accurately extending the Cramer-Rao lower bound of the mixing matrix to the Cramer-Rao lower bound of the ISR index. This technique avoids the overly optimistic performance bound caused by ignoring the influence of noise in traditional noiseless model analysis. It achieves an accurate characterization of the lower bound of residual interference power after blind source signal separation in noisy environments, thus providing a theoretically achievable performance limit for blind source signal separation communication anti-interference algorithms under noisy conditions. This solves the technical problems of unclear separation performance limits and lack of theoretical basis for algorithm design under noisy conditions.

[0020] In one embodiment, both the communication signal and the interference signal in the source signal vector are taken from the discrete modulation constellation point set, the mixing matrix is ​​determined according to the wireless channel model, and the noise is Gaussian white noise.

[0021] Specifically, the source signal vector in this application The noise vector is derived from the discrete modulation constellation point set generated by the modulation schemes of the communication signal and the interference signal, and follows a Gaussian distribution. The discrete constellation point model aligns with the characteristics of practical digital communication systems, and the Gaussian white noise assumption simplifies the derivation of the likelihood function. This modeling approach makes the calculated performance bounds practically meaningful.

[0022] In one embodiment, multiple sets of received signal vectors are constructed by: generating them through simulation according to preset communication signal modulation method, interference signal modulation method, noise distribution and wireless channel model; or, generating them by mixing them according to the actual collected communication signal, interference signal and noise signal according to the actual channel model.

[0023] Specifically, the received signal vector can be represented as ,in, This represents the source signal vector generated according to the modulation method of the communication signal and the interference signal. Represents a mixture matrix. This represents the set of source signal vectors composed of discrete modulation constellation points. Furthermore, multiple sets of received signal vectors can be constructed by mixing actual acquired communication signals, interference signals, and noise signals according to the actual channel model. Simulation generation facilitates the batch generation of multi-sample data and is suitable for Monte Carlo calculations; actual acquisition methods can reflect the non-ideal characteristics of the real environment, enhancing the engineering credibility of the performance metrics.

[0024] In one embodiment, calculating the partial derivative of the log-likelihood function with respect to the mixing matrix includes: The partial derivative of the log-likelihood function with respect to the mixture matrix is ​​calculated as follows:

[0025] in, Indicates the noise variance. This indicates that the matrix is ​​vectorized column-wise. Represents the conjugate transpose of a matrix. Represents a set The first in A vector, This represents the set of source signal vectors composed of discrete modulation constellation points. Represents the received signal vector. Represents a mixture matrix. , This represents the natural exponential function.

[0026] Specifically, this calculation method makes full use of the discrete characteristics of the source signal, which is the basis for the subsequent accurate estimation of the Fisher information matrix.

[0027] In one embodiment, the Monte Carlo estimate of the Fisher information matrix is:

[0028] in, , , Represents the transpose of a matrix. Represents the inverse of a matrix. Represents the conjugate of a matrix. It represents the partial derivative of the mixture matrix.

[0029] Specifically, the estimate is obtained by averaging the outer product of the partial derivatives of all received samples and dividing by the noise variance.

[0030] In one embodiment, the gain matrix is ​​defined based on the estimated mixing matrix and the ideal separation matrix. ,in, This represents the mixture matrix estimated by the separation algorithm. Represents the ideal separation matrix. This represents a mixture matrix.

[0031] Specifically, the ideal separation matrix satisfy When separation is ideal, Then the gain matrix (Identity matrix), where residual interference is zero. In practice, the existence of estimation errors makes... The non-zero off-diagonal elements directly reflect the residual power ratio of the interference.

[0032] In one embodiment, calculating the Jacobian matrix required to map from the mixing matrix to the gain matrix includes: The Jacobian matrix required to map from the mixing matrix to the gain matrix is ​​calculated as follows:

[0033] in, Indicates the Kronecker product. Indicates the first A column vector with 1 element and all other elements being zero. i In the above equation, 1 and 2, Represents the ideal separation matrix. This represents the transpose of a matrix.

[0034] Specifically, the Jacobian matrix A linear mapping relationship was established between small changes in the mixture matrix and small changes in the gain matrix. Using this matrix, the Cramer-Rao lower bound of the mixture matrix estimation can be transformed into the Cramer-Rao lower bound of the gain matrix (i.e., the ISR index).

[0035] In one embodiment, the Cramero lower bound of the power ratio of the residual interference signal to the communication signal is calculated using the inverse of the Fisher information matrix and the Jacobian matrix, including:

[0036] in, Represents the trace of a matrix. This represents the number of independent observation points used in the blind source signal separation algorithm. Represents the Jacobian matrix. Represents the transpose of a matrix. Represents the Fisher information matrix. Represents the inverse of a matrix. Denotes the Cramérault lower bound for the off-diagonal elements of the gain matrix. This represents the off-diagonal elements of the gain matrix.

[0037] Specifically, this formula gives the lower bound of ISR based on Cramer-Rowe's inequality. In the formula... The lower bound is the number of independent observation points, typically equal to the number of sets of received signal vectors or the number of effective samples after some transformation. This lower bound represents the minimum mean square error achievable by any unbiased estimator, and therefore can serve as a benchmark for evaluating the anti-interference performance of practical blind source signal separation algorithms.

[0038] In one embodiment, the method is applied to a scenario comprising a communication signal transmitter, an interference signal transmitter, and a dual-channel receiver terminal, wherein the receiver uses blind source signal separation technology to estimate the separation matrix by receiving the mixed signal, thereby separating the communication signal from the interference signal to achieve communication anti-interference.

[0039] Specifically, to verify the boundary analysis method for anti-interference performance of blind source signal separation communication provided in this embodiment, simulation experiments were conducted. For example... Figure 2 As shown, the parameters involved in the simulation experiment are as follows: both the communication signal and the interference signal are QPSK modulated, the signal-to-noise ratio is 10dB, and the mixing matrix... The signal was generated according to the Rayleigh fading model, with the Monte Carlo iteration set to 1e4. The JADE algorithm was used to separate the blind source signal. Figure 2 Simulation results show that, compared with the performance boundary analysis results of blind source signal separation under the existing noiseless model, the performance boundary analysis method proposed in this application takes into account the influence of noise, and the performance boundary analysis results are more accurate and closer to the performance obtained by the actual separation algorithm.

[0040] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0041] In one embodiment, such as Figure 3 As shown, a blind source signal separation communication anti-interference performance boundary analysis device is provided, comprising: The signal construction module 302 is used to construct multiple sets of received signal vectors. Each set of received signal vectors is obtained by mixing the source signal vectors through a mixing matrix and then adding noise. The partial derivative calculation module 304 is used to calculate the partial derivative of the log-likelihood function with respect to the mixing matrix for each set of received signal vectors. Information matrix estimation module 306 is used to take the statistical expectation of all partial derivatives as the Monte Carlo estimate of the Fisher information matrix; The mapping transformation module 308 is used to define the gain matrix based on the estimated mixing matrix and the ideal separation matrix, and to calculate the Jacobian matrix required to map from the mixing matrix to the gain matrix; The performance limit output module 310 is used to calculate the Cramer-Rao lower bound of the ratio of residual interference signal power to communication signal power using the inverse matrix of the Fisher information matrix and the Jacobian matrix, and outputs the Cramer-Rao lower bound as the performance limit of blind source signal separation communication anti-interference.

[0042] Specific limitations regarding the blind source signal separation communication anti-interference performance boundary analysis device can be found in the above-described limitations regarding the blind source signal separation communication anti-interference performance boundary analysis method, and will not be repeated here. Each module in the aforementioned blind source signal separation communication anti-interference performance boundary analysis device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of the processor in a computer device, or stored in software in the memory of a computer device, so that the processor can call and execute the operations corresponding to each module.

[0043] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0044] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for analyzing the anti-interference performance of blind source signal separation communication, characterized in that, The method includes: Multiple sets of received signal vectors are constructed. Each set of received signal vectors is obtained by mixing source signal vectors with a mixing matrix and then adding noise. The source signal vectors are selected from the discrete modulation constellation point set generated by the modulation method of the communication signal and the modulation method of the interference signal. For each set of received signal vectors, the partial derivative of the log-likelihood function with respect to the mixing matrix is ​​calculated as follows: in, Indicates the noise variance. This indicates that the matrix is ​​vectorized column-wise. Represents the conjugate transpose of a matrix. Represents a set The first in A vector, This represents the set of source signal vectors composed of discrete modulation constellation points. Represents the received signal vector. Represents a mixture matrix. , Represents the natural exponential function; The statistical expectation of all partial derivatives is used as the Monte Carlo estimate of the Fisher information matrix; the Monte Carlo estimate of the Fisher information matrix is: in, , , Represents the transpose of a matrix. Represents the inverse of a matrix. Represents the conjugate of a matrix. Represents the partial derivatives of the mixture matrix; Define the gain matrix based on the estimated mixing matrix and the ideal separation matrix, and calculate the Jacobian matrix required to map from the mixing matrix to the gain matrix; Using the inverse of the Fisher information matrix and the Jacobian matrix, the Cramer-Rao lower bound of the ratio of residual interference signal power to communication signal power is calculated, and the Cramer-Rao lower bound is output as the performance limit of blind source signal separation communication anti-interference.

2. The method according to claim 1, characterized in that, The communication signal and interference signal in the source signal vector are both taken from the discrete modulation constellation point set, the mixing matrix is ​​determined according to the wireless channel model, and the noise is Gaussian white noise.

3. The method according to claim 1, characterized in that, The multiple sets of received signal vectors are constructed by: generating them through simulation according to preset communication signal modulation methods, interference signal modulation methods, noise distribution, and wireless channel models; or, generating them by mixing them according to the actual collected communication signals, interference signals, and noise signals, based on the actual channel model.

4. The method according to claim 1, characterized in that, Based on the estimated mixture matrix and the ideal separation matrix, the gain matrix is ​​defined as follows: ,in, This represents the mixture matrix estimated by the separation algorithm. Represents the ideal separation matrix. This represents a mixture matrix.

5. The method according to claim 1, characterized in that, Calculating the Jacobian matrix required to map from the mixing matrix to the gain matrix includes: The Jacobian matrix required to map from the mixing matrix to the gain matrix is ​​calculated as follows: in, Indicates the Kronecker product. Indicates the first A column vector with 1 element and all other elements being zero. i For 1 and 2, Represents the ideal separation matrix. This represents the transpose of a matrix.

6. The method according to claim 1, characterized in that, Using the inverse of the Fisher information matrix and the Jacobian matrix, the Cramero lower bound of the power ratio of the residual interference signal to the communication signal is calculated, including: in, Represents the trace of a matrix. This represents the number of independent observation points used in the blind source signal separation algorithm. Represents the Jacobian matrix. To represent the transpose of a matrix, Represents the Fisher information matrix. Represents the inverse of a matrix. Denotes the Cramérault lower bound for the off-diagonal elements of the gain matrix. This represents the off-diagonal elements of the gain matrix.

7. The method according to claim 1, characterized in that, The method is applied to a scenario that includes a communication signal transmitter, an interference signal transmitter, and a dual-channel receiving terminal. The receiving terminal uses blind source signal separation technology to estimate the separation matrix by receiving the mixed signal, thereby separating the communication signal from the interference signal to achieve communication anti-interference.

8. A device for analyzing the anti-interference performance of blind source signal separation communication, characterized in that, The device includes: The signal construction module is used to construct multiple sets of received signal vectors. Each set of received signal vectors is obtained by mixing source signal vectors with a mixing matrix and then adding noise. The source signal vectors are selected from the discrete modulation constellation point set generated according to the modulation method of the communication signal and the modulation method of the interference signal. The partial derivative calculation module is used to calculate the partial derivative of the log-likelihood function with respect to the mixing matrix for each set of received signal vectors. in, Indicates the noise variance. This indicates that the matrix is ​​vectorized column-wise. Represents the conjugate transpose of a matrix. Represents a set The first in A vector, This represents the set of source signal vectors composed of discrete modulation constellation points. Represents the received signal vector. Represents a mixture matrix. , Represents the natural exponential function; The information matrix estimation module is used to obtain the Monte Carlo estimate of the Fisher information matrix from the statistical expectations of all partial derivatives; the Monte Carlo estimate of the Fisher information matrix is: in, , , To represent the transpose of a matrix, Represents the inverse of a matrix. Represents the conjugate of a matrix. Represents the partial derivatives of the mixture matrix; The mapping transformation module is used to define a gain matrix based on the estimated mixing matrix and the ideal separation matrix, and to calculate the Jacobian matrix required to map from the mixing matrix to the gain matrix; The performance limit output module is used to calculate the Cramer-Rao lower bound of the ratio of residual interference signal power to communication signal power using the inverse matrix of the Fisher information matrix and the Jacobian matrix, and output the Cramer-Rao lower bound as the performance limit of blind source signal separation communication anti-interference.

Images