A semi-supervised heterosignal unmixing correction method

Abstract

The application discloses a semi-supervised heterogenous signal demixing correction method. The method is used for solving the problem that the corresponding relationship between the antenna array receiving signal as a training sample and the known radiation source direction as a label is interfered by the aliasing signal. The method uses the antenna array receiving signal and the direction information of part of the heterogenous signal to realize the semi-supervised correction of the amplitude and phase response of the antenna array receiving signal. The method uses the direction information of part of the heterogenous signal to realize the semi-supervised correction of the amplitude and phase response of the antenna array receiving signal, and is suitable for the occasions that the heterogenous signal is time-frequency aliasing and the inconsistency error of the amplitude and phase response of the antenna array receiving signal needs to be compensated on line or in situ.

Description

Technical Field

[0001] This invention belongs to the field of passive detection technology, specifically relating to a semi-supervised method for demixing and correcting heterogeneous signals. Background Technology

[0002] In the field of passive detection technology, antenna arrays are commonly used to receive signals radiated from radiation sources. This not only improves signal detection performance by increasing the signal-to-noise ratio (SNR) gain of the antenna array, but also allows for estimation of the direction of arrival of the radiation source signal using array signal processing methods. The main principle of this commonly used direction-finding method is that the direction vector of the antenna array is consistent with the amplitude and phase response of the received signal. However, actual antenna array receiving channels have amplitude and phase response errors, which disrupt the consistency between the antenna array's direction vector and the amplitude and phase response of the received signal. This not only leads to a decrease in the antenna array's SNR gain but also increases the direction-finding error, and may even cause the direction-finding function to fail. Therefore, before signal detection or direction finding, it is necessary to correct the amplitude and phase response of the received signal, i.e., estimate the amplitude and phase response of the received signal to provide amplitude and phase compensation values ​​to minimize the inconsistency between the antenna array's direction vector and the amplitude and phase response of the received signal.

[0003] A common method for estimating the amplitude and phase response of an antenna array's received signal is to use the received signal and the known azimuth of the radiating source. This is achieved by dividing each element of the received signal vector by the corresponding element of the antenna array's direction vector at the azimuth of the radiating source. This method is essentially a supervised correction method, comprising a training phase and an operational phase. During the training phase, the received signal serves as the training sample, and the known azimuth of the radiating source acts as the label. The amplitude and phase response of the received signal are estimated using the training sample and the label. During the operational phase, the amplitude and phase response estimated during the training phase are used to compensate for the inconsistency error between the antenna array's direction vector and the amplitude and phase response of the received signal.

[0004] However, since the signals radiated by known radiation sources generally have finite bandwidth, the corresponding antenna array received signals are difficult to use to estimate the amplitude and phase responses of antenna array received signals in other frequency bands. Furthermore, when the antenna array received signals are aliased with signals radiated from other unknown sources with different directions of arrival, the correspondence between the antenna array received signals used as training samples and the known radiation source locations used as labels is interfered with by the aliasing signals, leading to a deterioration in the performance of this supervised amplitude and phase response estimation method, making it difficult to apply.

[0005] Since the azimuth of some radiation sources corresponding to the antenna array received signal is known after a certain period of passive detection within the passive detection frequency band, it is necessary to combine the training phase and the working phase into one. This involves developing a method that utilizes the azimuth information of some heterogeneous signals to perform semi-supervised correction on the amplitude and phase response of the antenna array received signal when the correspondence between the antenna array received signal (used as a training sample) and the known azimuth of the radiation source (used as a tag) is interfered with by aliasing signals. Summary of the Invention

[0006] This invention addresses the problem of aliasing interference in the correspondence between the antenna array received signal used as a training sample and the known radiation source location used as a tag. It utilizes the location information of the antenna array received signal and some heterogeneous source signals to achieve semi-supervised correction of the amplitude and phase response of the antenna array received signal.

[0007] A semi-supervised heterogeneous signal demixing correction method is proposed. First, the number of antennas in the antenna array, the number of radiation sources with known azimuths, the known azimuths of the radiation sources and their corresponding direction vectors, and the number of snapshots of the received signal from the antenna array are set. Second, the received signal from the antenna array and its sample autocorrelation matrix are determined. Then, singular value decomposition is performed on the sample autocorrelation matrix of the received signal to determine the number of small singular values ​​and the singular vectors corresponding to the small singular values. Next, the heterogeneous signal demixing matrix is ​​determined using the singular vectors corresponding to the small singular values ​​and the direction vectors corresponding to the known azimuths of the radiation sources. Finally, the complex response of the semi-supervised antenna array received signal is determined from the heterogeneous signal demixing matrix, thereby determining the amplitude response estimate and phase response estimate of the semi-supervised antenna array received signal.

[0008] The present invention specifically includes the following steps:

[0009] S1. Set the number of antennas in the antenna array to... The number of radiation sources with known orientations is: The known locations of the radiation sources are: The corresponding direction vector is The number of snapshots for the antenna array to receive signals is ;

[0010] S2. Determine the antenna array receiving signal as... 1-th order matrix The sample autocorrelation matrix is ​​as follows:

[0011] ;

[0012] in, This is the conjugate transpose of the matrix;

[0013] S3. The sample autocorrelation matrix of the signal received by the antenna array. Perform singular value decomposition to determine the number of small singular values. The singular vectors corresponding to small singular values ​​are , And determine the singular matrix as:

[0014] ;

[0015] S4. Utilizing the singular vectors corresponding to small singular values. , The direction vector corresponding to the known location of the radiation source The heterogeneous signal demixing matrix is ​​determined as follows:

[0016] ;

[0017] S5. Based on the heterogeneous signal demixing matrix, the complex response of the semi-supervised antenna array received signal is determined as follows:

[0018] ;

[0019] Therefore, the amplitude response estimate of the received signal from the antenna array is determined as follows: The phase response is estimated as ;in This represents the first column vector of the unmixing matrix. Represents the second to third elements of the unmixing matrix. A matrix consisting of column vectors. Representation matrix The generalized inverse matrix, This represents the magnitude of each complex element of the vector. Representing vectors The first element, This represents the phase of each complex element of the vector.

[0020] The beneficial effects of this invention are: by utilizing the azimuth information of some heterogeneous signals, the amplitude and phase response of the antenna array received signal are semi-supervised for correction, which is suitable for occasions where heterogeneous signals are mixed in time and frequency, and where online or on-site compensation is required for the inconsistency error of the amplitude and phase response of the antenna array received signal. Detailed Implementation

[0021] All features disclosed in this specification, or all steps in all disclosed methods or processes, may be combined in any way, except for mutually exclusive features and / or steps.

[0022] Any feature disclosed in this specification (including any appended claims and abstract) may be replaced by other equivalent or similar features, unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.

[0023] Example: In this example, the number of antennas in the antenna array is set. =8. The number of radiation sources with known orientations is The known locations of the radiation sources are: Spend, The degrees and their corresponding direction vectors are respectively The number of snapshots for the antenna array to receive signals is =64. In the antenna array receiving signal, the signal-to-noise ratio of the signal received by each antenna is 0dB. Not only are the signals of two radiation sources with known azimuth and time-frequency overlap superimposed, but also the signal of another radiation source with unknown azimuth and time-frequency overlap at 25.16 degrees is superimposed, causing the supervised correction method to fail.

[0024] Normalized with the first element of the antenna array as a reference, the actual complex response of the received signal from the antenna array is as follows:

[0025] 1.0000+0.0000i, 0.9416+0.3886i, -0.9226+0.7207i, 0.3811+1.0671i,

[0026] 0.8015-0.5922i, -0.5252-0.7259i, 0.8922-0.5698i, 0.2131-1.0072i;

[0027] Using the first element of the antenna array as a reference for normalization, and employing a semi-supervised heterogeneous signal demixing correction method of this invention, the determined complex response estimates are as follows:

[0028] 1.0000+0.0000i, 0.9516+0.3246i, -0.7621+0.7217i, 0.4441+0.9885i,

[0029] 0.6940-0.6564i, -0.5175-0.7214i, 0.7277-0.5298i, 0.1692-0.9649i.

[0030] A comparison of the actual complex response of the antenna array received signal with the complex response determined by the method of this invention shows that the amplitude response estimation errors of the method of this invention are 0.06, 0.47, 0.10, 0.18, 0.04, 0.70, and 0.22 (in dB), all less than 0.71 dB; the phase response estimation errors of the method of this invention are 3.59, 5.44, 4.54, 6.95, -0.23, 3.49, and 2.00 (in degrees), all less than 7.00 degrees. Therefore, by using the method of this invention, even when the correspondence between the antenna array received signal (used as a training sample) and the known azimuth of the radiation source (used as a tag) is interfered with by aliasing signals, the method can utilize the azimuth information of the antenna array received signal and some heterogeneous signals to achieve semi-supervised correction of the amplitude and phase response of the antenna array received signal.

[0031] This invention is not limited to the specific embodiments described above. The invention extends to any new feature or combination disclosed in this specification, as well as any new method or process step or combination disclosed herein.

Claims

1. A semi-supervised method for demixing and correcting heterogeneous signals, characterized in that... Includes the following steps: S1: Set the number of antennas in the antenna array, the number of radiation sources with known azimuths, the known azimuths of the radiation sources and their corresponding direction vectors, and the number of snapshots of the antenna array receiving signals; S2: Determine the received signal of the antenna array and its sample autocorrelation matrix. for: ; S3: Perform singular value decomposition on the sample autocorrelation matrix of the received signal from the antenna array to determine the number of small singular values ​​and the corresponding singular vectors, and the singular matrix. for: ; S4: Determine the unmixing matrix of the heterogeneous signal using the singular vectors corresponding to the small singular values ​​and the direction vectors corresponding to the known radiation source locations. for: ； S5: Determine the complex response of the received signal from the semi-supervised antenna array using the heterogeneous signal demixing matrix. for: This allows for the determination of the amplitude response estimate of the semi-supervised antenna array received signal. and phase response estimation ; in: This is a multi-order matrix for the antenna array to receive signals. This is the conjugate transpose of the matrix. For the singular vectors corresponding to the smaller singular values, , This is the direction vector corresponding to the known location of the radiation source. , This represents the first column vector of the unmixing matrix. Represents the second to third elements of the unmixing matrix. A matrix consisting of column vectors. Representation matrix The generalized inverse matrix, This represents the magnitude of each complex element of the vector. Representing vectors The first element, This represents the phase of each complex element of the vector.

Images