Satellite navigation anti-interference array variable polarization interference detection and discrimination method and device

Abstract

This invention relates to the field of satellite navigation anti-jamming technology, and discloses a method and apparatus for detecting and discriminating variable polarization interference in satellite navigation anti-jamming arrays. The method acquires a snapshot vector sequence of a single-polarization antenna array, estimates the covariance matrix using short and long time windows in parallel, and extracts the rotation angular rate, rotation direction consistency index, effective rank of the short time window, and effective rank of the long time window from the principal eigenvectors. Based on this, it determines whether the interference is a single fixed polarization interference, two fixed polarization interferences, a rotating variable polarization single interference source, or a switching variable polarization single interference source. When it is determined to be a rotating variable polarization single interference source, the estimated value of the polarization change rate is output; when it is determined to be a switching variable polarization single interference source, the estimated value of the switching period is output, and the interference situation category and its corresponding dynamic characteristic parameters are provided to the subsequent CRPA anti-jamming processing chain. This invention does not require changes to existing single-polarization CRPA hardware and can improve the accuracy of variable polarization interference situation discrimination.

Description

Technical Field

[0001] This invention belongs to the field of satellite navigation anti-interference technology, specifically relating to a method and device for detecting and judging polarization interference in satellite navigation anti-interference arrays. Background Technology

[0002] Satellite navigation receivers in high-reliability applications are typically equipped with a single-polarization controlled radiation pattern antenna array (CRPA). This type of array consists of... Composed of single-polarized array elements with the same expected polarization response, a pattern nulling against the direction of interference is formed through adaptive weights; for The single-polarization CRPA of the array element, whose degrees of freedom can be used for anti-interference zeroing, is usually limited. constraint.

[0003] In a single-polarization CRPA receiving scenario, a variable-polarization interference source, through a dual-polarization transmit antenna and amplitude, phase, and timing control, causes the polarization state of the interference signal to change over time. Since single-polarization array elements still exhibit non-zero responses to cross-polarization components, and the cross-polarization responses of different array elements are not entirely consistent, a single variable-polarization interference source may form two independent equivalent spatial components within a long time window. After eigenvalue decomposition of the long-time-window covariance matrix, a set of eigenvalues ​​arranged in numerical order constitutes the eigenvalue spectrum of the long-time-window covariance matrix; among them, eigenvalues ​​exceeding the noise floor threshold and capable of characterizing the effective interference subspace dimension are called large eigenvalues. Based on the above mechanism, a single variable-polarization interference source may exhibit two large eigenvalues ​​in the eigenvalue spectrum of the long-time-window covariance matrix; similarly, two fixed-polarization interference sources from different directions typically also exhibit two large eigenvalues ​​in the eigenvalue spectrum of the long-time-window covariance matrix. Therefore, based solely on the static eigenvalue spectrum of the long-term window covariance matrix, a single variable polarization interference source and two fixed polarization interference sources may appear to be isomorphic at the numerical level.

[0004] To address the aforementioned eigenvalue spectrum isomorphism problem, existing processing methods mainly include: polarization-based interference processing methods in the radar field, interference type identification methods in satellite navigation receivers, and interference source number estimation methods in array received data. Among these, polarization-based interference processing methods in the radar field typically rely on dual-polarization or fully polarized receiving channels and segmented estimation based on pulse structure or polarization stability segments, making them difficult to directly apply to continuous wave reception scenarios in single-polarization CRPA satellite navigation receivers. Interference identification methods in satellite navigation receivers primarily target broadband interference, narrowband interference, pulse interference, frequency sweeping interference, or different modulation types, and generally do not consider whether the polarization state changes over time as an independent interference attribute. Interference source number estimation methods in array received data mainly estimate the number of effective interference sources based on the static eigenvalue spectrum of the covariance matrix. However, when a single polarization-based interference source and two fixed-polarization interference sources both exhibit two large eigenvalues, it is difficult to distinguish between these two types of interference scenarios solely based on the static eigenvalue spectrum.

[0005] Under the aforementioned static eigenvalue spectrum isomorphism, judging the interference situation solely based on the eigenvalue spectrum of the long-term window covariance matrix can easily lead to erroneous prior information about the interference in the subsequent CRPA anti-interference processing chain. Specifically, if one variable-polarity interference source is misclassified as two fixed-polarity interference sources, the subsequent adaptive zeroing algorithm will perform null calculations based on two independent interference directions. However, these two independent interference directions do not actually exist, which may cause the null position to drift with changes in the interference polarization state, resulting in repeated adjustments to the adaptive weights. Conversely, if two fixed-polarity interference sources are misclassified as one variable-polarity interference source, the subsequent adaptive zeroing algorithm may only maintain nulls for one equivalent interference direction, thus missing the null corresponding to the other real fixed-polarity interference source. Furthermore, existing solutions typically only output the interference type or the number of interference sources, without simultaneously outputting dynamic feature parameters such as polarization change rate and switching cycle. This makes it difficult to provide parameter basis for the selection of window length, sample selection strategy, and weight refresh cycle in the subsequent CRPA anti-interference processing chain. Summary of the Invention

[0006] The technical problem to be solved by this invention is: how to accurately distinguish between fixed polarization interference and variable polarization interference of different forms without changing the existing single-polarization CRPA hardware structure, and obtain dynamic characteristic parameters that can be used for subsequent anti-interference processing.

[0007] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: Firstly, a method for detecting and discriminating variable polarization interference in satellite navigation anti-jamming arrays is proposed, including: right The received signals from each single-polarization array element are down-converted and sampled to obtain a snapshot vector sequence. It is an integer not less than 4; The sample covariance of the snapshot vector sequence is estimated in parallel using short and long time windows, resulting in the short-window covariance matrix and the long-window covariance matrix, with the long window length being greater than the short window length. Eigenvalue decomposition is performed on the two covariance matrices respectively, and the principal eigenvector of the unit norm corresponding to the largest eigenvalue of the short time window and the eigenvectors corresponding to the two largest eigenvalues ​​of the long time window are extracted. The long time window eigenvectors span a two-dimensional interference subspace. The rotation angular rate is calculated based on the angle between adjacent principal feature vectors and the time interval between the centers of short time windows; the principal feature vectors are projected onto a continuously aligned two-dimensional interference subspace, and the rotation direction consistency index is calculated based on the surround correction angle increment of the projection angle sequence. The threshold is determined by using the product of the noise power estimate and the threshold coefficient as the largest eigenvalue. The number of eigenvalues ​​exceeding the threshold in the two covariance matrices is counted to obtain the effective rank of the short window and the effective rank of the long window, respectively. The interference situation category is determined based on the rotation angular rate, rotation direction consistency index, and effective rank of short and long time windows, including single fixed polarization interference, two fixed polarization interference, rotating variable polarization single interference source, and switching variable polarization single interference source. When determined to be a rotating variable polarization single interference source, output the estimated polarization change rate; when determined to be a switching variable polarization single interference source, and the stable segments before and after the candidate switching event and between adjacent events can be classified into two discrete polarization states. , Furthermore, when the status labels alternate over time, the estimated switching cycle value is output, and the switching cycle parameter is kept in a pending update state; the interference situation category and the confirmed valid dynamic characteristic parameters are used as adaptive trigger parameters for the CRPA anti-interference processing chain.

[0008] Secondly, a satellite navigation anti-jamming array polarization interference detection and discrimination device is proposed, comprising: Array element receiving unit, used for... The received signals from each single-polarization array element are down-converted and sampled to output a snapshot vector sequence. It is an integer not less than 4; The dual-time-scale covariance estimation unit is used to estimate the sample covariance of the snapshot vector sequence in parallel using a short time window and a long time window, and output the short and long time window covariance matrices, with the long time window being longer than the short time window. The main feature analysis unit is used to perform eigenvalue decomposition on the two covariance matrices and outputs the unit norm principal eigenvector corresponding to the largest eigenvalue of the short time window, the two-dimensional interference subspace spanned by the eigenvectors corresponding to the two largest eigenvalues ​​of the long time window, and the eigenvalue set of the two matrices. The rotation angular rate and consistency index calculation unit is used to calculate the rotation angular rate based on the angle between adjacent principal feature vectors and the time interval between the centers of short time windows, and to project the principal feature vectors onto a continuously aligned two-dimensional interference subspace, and to calculate the rotation direction consistency index based on the surround correction angle increment of the projection angle sequence. The effective rank calculation unit is used to count the number of eigenvalues ​​in two matrices that exceed the threshold by using the product of the noise power estimate and the threshold coefficient as the threshold, and output the effective rank of the short window and the effective rank of the long window. The joint decision unit is used to determine the interference situation category based on the rotation angular rate, rotation direction consistency index, short window effective rank and long window effective rank, and output the category and effective dynamic characteristic parameters. The categories include single fixed polarization interference, two fixed polarization interference, rotating variable polarization single interference source and switching variable polarization single interference source. The decision result output interface is used to output the interference situation category and dynamic characteristic parameters to the subsequent CRPA anti-interference processing unit.

[0009] Compared with existing technologies, this invention has the following advantages and beneficial effects: By estimating the sample covariance of the snapshot vector sequence of a single-polarization antenna array using short-time and long-time windows respectively, the spatial characteristic differences exhibited by a variable-polarization single interference source at different time scales provide a basis for distinguishing it from a fixed-polarization interference source. Specifically, the principal eigenvectors of the short-time window covariance matrix are used to characterize the change in the instantaneous equivalent spatial direction of the interference source. The rotation angular rate is calculated using adjacent principal eigenvectors, and the principal eigenvectors are projected onto the two-dimensional interference subspace corresponding to the long-time window covariance matrix to further obtain a rotation direction consistency index reflecting the continuous rotation direction. Simultaneously, by statistically analyzing the number of eigenvalues ​​exceeding the large eigenvalue judgment threshold in both the short-time and long-time window covariance matrices, the effective rank of the short-time window and the effective rank of the long-time window are obtained, thus characterizing the dimensional changes of the interference subspace at different time scales. Therefore, by jointly judging the rotation angular rate, rotation direction consistency index, effective rank of the short window, and effective rank of the long window, it is possible to distinguish between single fixed-polarization interference, two fixed-polarization interference, rotating variable-polarization single interference sources, and switching variable-polarization single interference sources, reducing the risk of misjudgment of the interference situation caused by relying solely on the static eigenvalue spectrum of the long window. When it is determined to be a rotating or switching variable-polarization single interference source, the estimated polarization change rate or switching period is further output, and the interference situation category and its corresponding dynamic characteristic parameters are provided to the subsequent CRPA anti-interference processing chain, providing a basis for the adaptive adjustment of subsequent sample selection, weight update, and null maintenance. Attached Figure Description

[0010] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram illustrating the response relationship between a single-polarization array element and co-polarization and cross-polarization components provided in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram of the overall process of the satellite navigation anti-jamming array polarization interference detection and discrimination method provided in Embodiment 1 of the present invention; Figure 3 This is a schematic diagram showing the trajectory of the principal feature vector on the two-dimensional interference subspace plane provided in Embodiment 1 of the present invention. Figure 4 This is a schematic diagram showing the comparison between dual-time-scale covariance estimation and effective rank response under different disturbance conditions provided in Embodiment 1 of the present invention; Figure 5 This is a schematic diagram of the joint decision state transition provided in Embodiment 1 of the present invention; Figure 6 This is a structural block diagram of the satellite navigation anti-interference array polarization interference detection and discrimination device provided in Embodiment 2 of the present invention.

[0011] The attached diagram shows the markings and corresponding component names: 100 - Array element receiving unit; 101 - Single-polarized array element; 102 - RF front-end; 103 - Down-conversion and sampling channel; 200 - Dual-time-scale covariance estimation unit; 201 - Short-time-window covariance estimation subunit; 202 - Long-time-window covariance estimation subunit; 300 - Principal feature analysis unit; 301 - Short-time-window eigenvalue decomposition subunit; 302 - Long-time-window eigenvalue decomposition subunit; 400 - Rotation angular rate and consistency index calculation unit; 401 - Rotation angular rate calculation subunit; 402 - Rotation direction consistency index calculation subunit; 500 - Effective rank calculation unit; 501 - Short-time-window effective rank counting subunit; 502 - Long-time-window effective rank counting subunit; 600 - Joint decision unit; 601 - Decision table matching subunit; 602 - Hysteresis decision subunit; 700 - Decision result output interface; 800 - Subsequent CRPA anti-interference processing unit. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. The illustrative embodiments and descriptions of this invention are for illustrative purposes only and are not intended to limit the invention. The embodiments described below are some, but not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0013] Example 1: Addressing the issue that in single-polarization CRPA satellite navigation reception scenarios, a single variable-polarization interference source and two fixed-polarization interference sources may exhibit isomorphism in the static eigenvalue spectrum of the long-time-window covariance matrix, making it difficult to distinguish the aforementioned interference patterns based solely on the static eigenvalue spectrum, this example provides a method for detecting and discriminating variable-polarization interference in satellite navigation anti-jamming arrays. This method is executed in the embedded processing platform of a satellite navigation receiver; the satellite navigation receiver is equipped with a single-polarization controlled radiation pattern antenna array, wherein the antenna array consists of… It consists of single-polarization array elements with the same expected polarization response. It is a positive integer not less than 4. Before proceeding with the implementation steps of this method, the physical mechanism upon which this method is based will be explained as follows.

[0014] Suppose the interference signal received by the satellite navigation single-polarized antenna array originates from the following direction: The interference signal contains two orthogonal polarization components. and The response vectors of the array to the same polarization component and the cross polarization component are denoted as follows: and The array receiving snapshot can be represented as: .in, for Complex-dimensional snapshot vector, and For the complex interference envelopes in the two polarization directions, and for A complex response vector. Under fixed polarization, ,in It is a complex proportionality coefficient that does not change with time, therefore A single fixed-polarization interference source contributes rank 1 to the sample covariance matrix. Under varying polarization conditions, No longer maintaining a fixed ratio; when and When linearly independent and the covariance observation time span covers significant changes in the relative amplitude or relative phase of the two polarization components, the effective rank contribution of a single variable polarization interference source to the sample covariance matrix can be expressed as 2.

[0015] See Figure 1 It shows the response relationship of a satellite navigation single-polarized antenna array to co-polarized and cross-polarized components. Figure 1 In the future, it will be The interference signal is decomposed into two orthogonal polarization components. and The response vector of a satellite navigation single-polarized antenna array to its co-polarized components is... The response vector to the cross-polarization component is ,and and The array manifolds are not parallel to each other at the manifold level. Under fixed polarization conditions, and Maintaining a fixed complex proportional relationship, the two response vectors are combined into a fixed equivalent steering vector; under varying polarization conditions, and The relative amplitude or phase of the signal changes over time, causing the received signal to... and The variations within the resulting two-dimensional space explain the physical mechanism by which a variable polarization interference source behaves as a single instantaneous equivalent steering vector on a short timescale, while forming a two-dimensional interference subspace on a long timescale.

[0016] Based on the aforementioned physical mechanisms, this method divides the time span of covariance sample estimation into short-term and long-term windows. The short-term window, shorter than the expected polarization dwell time or expected polarization change cycle, is used to obtain the instantaneous equivalent spatial characteristics of the variable polarization interference source. When the polarization state change amplitude within the short-term window is less than a preset short-term stability threshold, the instantaneous equivalent steering vector can be considered stable, and the effective rank contribution of the variable polarization interference source to the short-term window covariance matrix is ​​1. When the polarization change rate is high, causing the short-term window to cover polarization states sufficient to form two linearly independent equivalent spatial components, the effective rank of the short-term window is 2. The long-term window covers multiple polarization change cycles or multiple polarization switching states, used to average the two-dimensional interference subspace swept by the variable polarization interference source, making its effective rank contribution to the long-term window covariance matrix 2. The spatial characteristics and effective rank differences at these different time scales constitute the information channel for distinguishing between variable polarization interference sources and fixed polarization interference sources.

[0017] Based on this, this method constructs two sets of complementary decision observations. The first set of decision observations focuses on the change of the principal characteristic direction over time, including the rotation angular rate of the principal characteristic vector and the rotation direction consistency index corresponding to the projection angle sequence of the principal characteristic vector in the two-dimensional interference subspace. This is used to characterize the dynamic characteristics of a rotating variable polarization single interference source that changes continuously along a single direction. The second set of decision observations includes the effective rank of a short time window and the effective rank of a long time window. This is used to characterize the effective rank differences of a switching variable polarization single interference source at different time scales and the steady-state characteristics of the effective rank corresponding to fixed polarization interference. By combining two sets of decision observations into a joint decision observation, and employing unmatched state processing and hysteresis decision, this method distinguishes between single fixed-polarization interference, two fixed-polarization interference, rotating variable-polarization single interference source, and switching variable-polarization single interference source. When the source is determined to be a rotating variable-polarization single interference source, the estimated value of the polarization change rate is output; when the source is determined to be a switching variable-polarization single interference source, the estimated value of the switching period is output. The final interference situation category and corresponding dynamic characteristic parameters are used as adaptive triggering parameters for the subsequent CRPA anti-interference processing chain.

[0018] Below, in conjunction with Figures 2 to 5 The specific implementation steps of this method are explained in detail below: like Figure 2 As shown, the method for detecting and discriminating polarization interference in satellite navigation anti-jamming arrays includes the following steps: Step 1: For satellite navigation single-polarized antenna arrays The received signals of each single-polarization array element are down-converted and sampled to obtain a snapshot vector sequence.

[0019] Satellite navigation single-polarized antenna array Each single-polarized array element receives spatial electromagnetic signals and obtains... RF signal reception; for The received RF signals are processed by RF front-end filtering and amplification, and then down-converted and sampled to obtain... Digital baseband signal; The digital samples of the digital baseband signal at the same sampling time are combined into one. A complex vector is used to obtain the snapshot vector corresponding to the sampling time; the snapshot vectors obtained at different sampling times are combined in chronological order to form a snapshot vector sequence.

[0020] The first in the snapshot vector sequence The snapshot vector corresponding to each sampling time is denoted as . , for 3D complex column vector; The first in The component corresponds to the first The single-polar array element in the first... The down-conversion and sampling output at each sampling time, where... .

[0021] Step 2: Estimate the short-time window sample covariance matrix corresponding to the snapshot vector sequence based on the short-time window length; estimate the long-time window sample covariance matrix corresponding to the snapshot vector sequence based on the long-time window length.

[0022] In this step, the short-term window length is recorded as... Long-term window duration is recorded as ,and .by Indicates the snapshot sampling sequence number, in This indicates the covariance update sequence number, and sets the number of... The snapshot sampling endpoint corresponding to the next covariance update is denoted as... When estimating the short-time window sample covariance matrix, take... to of A snapshot vector, according to Calculate the short-time window covariance matrix. Wherein, for A complex Hermitian positive semi-definite matrix. For the first A snapshot vector, This represents the conjugate transpose. The short-time window length satisfies... ,in The dimensionless sampling redundancy coefficient is between 5 and 10.

[0023] Furthermore, let the sampling interval between adjacent snapshots be... The short time window duration is Expected polarization dwell time and expected polarization change cycle The polarization change timescale is determined by the technical parameters of the interference source, calibration test data, or historical observation data. When the corresponding prior value is temporarily unavailable, the polarization change timescale is estimated using the pre-sampling autocorrelation method described below. The short-time window length satisfies... or ,in .

[0024] During presampling, first take and order ,in, This represents the center time interval of the short time window between adjacent covariance update times. The presampling coverage coefficient satisfies... , can be taken as 10; not less than The unit norm principal eigenvector is obtained at each covariance update time. , ,and This ensures that the lag search interval is not empty. Let and order ,according to Calculate the normalized autocorrelation; when the denominator is zero, do not output the autocorrelation time scale estimate. The search range for the lag is... The first one that satisfies the search criteria will be... The local peak position that is not less than the correlation coefficient of adjacent available hysteresis quantities is denoted as . and will This is denoted as a candidate value for the time scale of polarization change. A significant peak threshold is selected. When no significant peak is found, the time scale is not estimated using presampled data, and the current results that depend on unknown time coverage conditions are left as pending or unmatched. and These represent rounding down and rounding up, respectively.

[0025] The This serves only as a conservative time limit for the initial selection phase of a short time window, and is used for verification. It is not directly equivalent to the expected polarization residence time. Or the expected polarization change cycle For a switching-type variable polarization single interference source, the stable segment is obtained according to the dual-state stable segment confirmation method described in step 6. After that, with As an estimate of length of stay, and based on state ,state Alternating confirmation of complete same-state regression period as For rotating variable polarization single interference sources, The results are determined by the observations of a complete rotational regression, based on the technical parameters of the interference source, calibration tests, historical observations, or the principal characteristic direction. or If the corresponding interference type has not yet been confirmed, the result based on the corresponding time coverage condition remains in a pending or unmatched state.

[0026] Sampling redundancy coefficient From the candidate set Selected from the options, the short-term window length meets the requirements. For each candidate Pick Candidate values ​​that satisfy the time length constraint are screened, and the candidate value with the largest value is selected to improve the stability of covariance estimation within the allowable time resolution. Under static benchmark interference conditions, the jump measure of adjacent unit norm principal eigenvectors is calculated using the continuous update results within the candidate short-time window coverage interval, and the sample's... The quantile was determined as the short-time stability threshold. ,in Only the jump metric within the candidate short-time window coverage interval is retained, and it does not exceed [a certain value]. The length of the candidate window. It is only used for stability screening of short time window lengths, and is different from the static determination of adjacent update times in step 6 during the running phase. Sample and calibrate separately, without taking the same values. If no candidate value simultaneously meets the lower limit of sample size and time length constraints, maintain the pending state and increase the sampling rate or shorten the covariance update time interval before reselecting.

[0027] When estimating the long-time window sample covariance matrix, in the first... The snapshot sampling endpoint corresponding to the sub-covariance update At this point, take a length of The snapshot vector within the sliding window, according to Calculate the long-term window covariance matrix. for A complex Hermitian positive semi-definite matrix. Long-term window length. Covering multiple polarization change cycles or multiple polarization switching states; assuming a preset cycle coverage number of... ,but ,in Furthermore, from the candidate integer window lengths that satisfy the time coverage condition, platform storage space constraints, and the fact that the processing time of a single update does not exceed the update interval, the candidate window length with the smallest value is selected as the [window length]. If no candidate value satisfies all constraints, the pending state is maintained, and storage space is increased, the update interval is extended, or the periodic coverage is reduced, but the adjusted... Still satisfied .

[0028] It should be noted that... The configuration is set to cover multiple polarization change cycles or multiple polarization switching states. The purpose is to ensure that, within the corresponding time span, the two-dimensional interference subspace swept by the instantaneous equivalent steering vector of the variable polarization interference source is averaged into the long-term window covariance matrix. For a switching-type variable polarization single interference source, the two different polarization states are denoted as states. and state Long-term window at least covers the state ,state and two adjacent states or two adjacent states A complete switching cycle between; when the observed state switching is insufficient to meet the above coverage conditions, the current result remains in the pending or unmatched state.

[0029] It should also be noted that the time intervals between two consecutive covariance updates... The time interval between them can be less than the short-time window length. The corresponding time span refers to the sliding overlap between adjacent short-term windows. For example, the overlap ratio between adjacent short-term windows is set between 50% and 75%, which is used to improve the observation refresh rate in subsequent steps 3, 4, and 5 while ensuring the stability of the short-term window covariance matrix estimation. Furthermore, both the short-term and long-term window covariance matrices can be estimated using the time averaging of snapshot vectors within the sliding window, or they can be estimated using a recursive update method with an exponential forgetting factor. The exponential forgetting factor is calculated based on the correspondence between the equivalent number of samples and the short-term or long-term window length.

[0030] The short-term window covariance matrix and long-term window covariance matrix obtained in step 2 are used as inputs for subsequent steps 3 and 4.

[0031] Step 3: Perform eigenvalue decomposition on the short-time window covariance matrix to obtain the principal eigenvector corresponding to the eigenvalue with the largest value; calculate the rotation rate of the principal eigenvector based on the angle between the principal eigenvectors of the unit norm corresponding to adjacent short-time windows and the time interval between the centers of adjacent short-time windows.

[0032] The short-time window covariance matrix obtained in step 2 Perform eigenvalue decomposition to obtain There are eigenvalues ​​and their corresponding eigenvectors. The eigenvalues ​​are arranged in descending order of value, and the eigenvector corresponding to the largest eigenvalue is taken as the principal eigenvector. This eigenvector is then normalized to its unit norm, denoted as . ; for A dimensional complex column vector is used to represent the 3rd-order complex column vector. The second covariance update corresponds to the estimation of the instantaneous equivalent spatial direction of disturbances within a short time window.

[0033] For the Second and third The unit norm principal eigenvector corresponding to the covariance update and ,according to Calculate the angle between adjacent principal feature directions, and follow the... Calculate the rotational angular rate. Wherein, , The dimension of is radians per second. The time interval between the center of the short time window corresponding to two adjacent covariance updates is used; the inner product modulus is taken and truncated by 1 to eliminate the influence of arbitrary global complex phase and to avoid the input of the inverse cosine function exceeding the domain due to finite precision operations.

[0034] For a rotating variable polarization single interference source, its instantaneous equivalent steering vector changes continuously within the two-dimensional interference subspace. Therefore, the angle between adjacent principal characteristic directions... and rotational angular rate Typically, this exceeds the baseline jitter range caused by noise and estimated fluctuations; for fixed polarization interference sources, the principal characteristic direction is approximately stable over time. It falls within the preset baseline jitter range. The direction of rotation is a non-negative rate of change, and the rotation direction is characterized by the rotation direction consistency index obtained in step 4. The mean value within the dynamic feature observation window is used as an estimate of the polarization change rate.

[0035] It should be noted that the eigenvalue decomposition of the short-window covariance matrix can be performed using the complete eigenvalue decomposition method, the incremental eigenvalue decomposition method, or the subspace tracking method, to reduce the amount of redundant computation between adjacent covariance update times. Regardless of the method used, the obtained principal eigenvectors are normalized to the unit norm; the rotation angular rate is calculated using the modulus of the complex inner product of the normalized principal eigenvectors, and therefore is not affected by the sign flipping of the principal eigenvectors or arbitrary global complex phase.

[0036] The rotational angular rate obtained in step 3 This will be one of the observations used in the joint decision-making process in subsequent step 6.

[0037] Step 4: Perform eigenvalue decomposition on the long-term window covariance matrix to obtain the two eigenvectors corresponding to the two largest eigenvalues; construct a two-dimensional interference subspace based on the two eigenvectors; project the main eigenvectors onto the two-dimensional interference subspace to obtain the projection angle sequence, and calculate the rotation direction consistency index based on the direction consistency of the angle increments at adjacent times in the projection angle sequence.

[0038] The long-term window covariance matrix obtained in step 2 Perform eigenvalue decomposition, sort the eigenvalues ​​in descending order of value, and denote the two eigenvalues ​​with the largest values ​​as eigenvalues. and The corresponding original unit norm eigenvector is and and order The two feature vectors span the first... The two-dimensional disturbance subspace corresponding to the second covariance update.

[0039] To avoid discontinuities in projection angles caused by eigenvalue sorting and swapping, arbitrary complex phases of eigenvectors, and near-simultaneous eigenvalues, the eigenvalue interval index is first calculated. ,in, To match the lower limit of the numerical protection against receiver quantization noise, the degradation judgment threshold satisfies... It is acceptable During the initial update, the component with the largest magnitude in each original basis vector is used as the phase reference. This component is then adjusted to a non-negative real number, and the resulting basis matrix is ​​set to... From the second update onwards, ,right Perform singular value decomposition, where, and for 3D unitary matrix Composed of singular values A non-negative diagonal matrix; let , This indicates taking the minimum singular value, therefore .when When this occurs, it indicates that at least one two-dimensional direction lacks reliable overlap, and the current rotation direction consistency observation is recorded as invalid. As the lower limit of singular value protection, it can be taken as... .when and When, choose to make Largest permutation If any of the selected relevant moduli is not greater than If the current rotation direction consistency observation is invalid, then the observation is marked as invalid; otherwise, it is handled according to... Perform relevant matching and phase alignment. When and At that time, according to Align the two-dimensional basis matrix globally, where, .

[0040] After completing the relevant matching and phase alignment or the overall alignment of the two-dimensional basis matrix, according to Calculate alignment residuals; when When this happens, the current rotation direction consistency observation is recorded as invalid. Represents the Frobenius norm; the aligned residual threshold satisfies Under stable reference disturbance conditions, assume the reference alignment residual sample is... quantiles are ,Pick ,in, and demand the proceeds exemplarily take The minimum singularity criterion is used to confirm that both directions of the two-dimensional subspace have reliable overlap, and the alignment residual criterion is used to confirm that the aligned basis matrix has temporal continuity; only after both are passed is the projection angle calculated using the current aligned basis.

[0041] When continuity alignment is effective, the unit norm principal eigenvector obtained in step 3 will be... Projected onto a continuously aligned two-dimensional disturbance subspace, let , The projected energy is and order .when or When this happens, the current rotation direction consistency observation is recorded as invalid, and the current projection angle is not calculated. and As the lower limit of numerical protection, any value can be taken as... Otherwise, according to Calculate the projection angle. Indicates according to , The sign of the arctangent function in the fourth quadrant determines the quadrant. Indicates taking the real part, Denotes complex conjugation, and .

[0042] After obtaining the effective projection angle sequence, first calculate the difference between adjacent projection angles. ;when Repeat execution ,when Repeat execution until and the corrected Recorded as Then according to Calculate the rotation direction consistency index. Among them, ; The number of projected angles contained within the directional consistency observation window, and The observation window is Its corresponding One adjacent angle increment; This is a sign function; it outputs 1 when the input is positive, -1 when the input is negative, and 0 when the input is 0. It only outputs 1 when the input is positive, -1 when the input is negative, and 0 when the input is 0. Calculation when all projection angles and their adjacent increments are valid Otherwise, the current This is recorded as an invalid value. In one specific implementation, .

[0043] Furthermore, The physical meaning is: when the principal eigenvector rotates continuously in a single direction on the two-dimensional interference subspace plane, exist Within the corresponding observation window, the signs remain the same, and the absolute value of the mean of the sign sequence is close to 1; when the principal eigenvector exhibits a random back-and-forth oscillation on the two-dimensional disturbance subspace plane, exist The signs within the corresponding observation window alternate between positive and negative, and the absolute value of the mean of the sign sequence is close to 0. Therefore, The rotational direction continuity of the principal eigenvector in the two-dimensional disturbance subspace plane can be measured to eliminate directional ambiguity that exists when only the rotational angular rate is used.

[0044] See Figure 3 This diagram illustrates three typical projection trajectories of the principal eigenvector on the two-dimensional interference subspace plane corresponding to the long-term window covariance matrix. Under a fixed polarization interference situation, the projection position of the principal eigenvector remains within the range defined by a preset static judgment margin, and the projection trajectory appears as a stationary point or a locally jittering trajectory formed around the stationary point. Under a rotating variable polarization single interference source situation, the projection position of the principal eigenvector changes continuously in the same direction with each covariance update, and the projection trajectory appears as a continuous arc. In the switching variable polarization single interference source situation of this embodiment, the projection position of the principal eigenvector remains in two discrete polarization states. and The corresponding position, and when the polarization state changes, jumps from one position to another, and the projected trajectory is shown as a step change between two stationary points. Figure 3 The three trajectories shown are used to illustrate the dynamic feature differences corresponding to the projection angle sequence, rotation direction consistency index, and principal feature vector jump metric.

[0045] The consistency index of the projection angle sequence and rotation direction obtained in step 4 This is also one of the observations used in the subsequent joint decision-making process in step 6.

[0046] Step 5: Using the product of the noise power estimate and the threshold coefficient as the large eigenvalue decision threshold, count the number of eigenvalues ​​exceeding the large eigenvalue decision threshold in the short-term window covariance matrix and the long-term window covariance matrix respectively, and obtain the effective rank of the short-term window and the effective rank of the long-term window.

[0047] First, according to Determine the large eigenvalue decision threshold, where The threshold coefficient, For noise power estimation. The number of samples in the long time window reaches [number missing]. Then, the eigenvalues ​​of the long-time window covariance matrix are arranged in descending order of value, and the dimension of the effective signal subspace is estimated using the minimum description length criterion or the Akaike information content criterion. , will the The first to the second The eigenvalues ​​are determined to be a non-empty set of small eigenvalues. and in accordance with Calculate noise power estimation; when When this happens, the current observation window is recorded as an unmatched state, where This is a preset noise power protection lower limit based on the receiver's quantization noise floor. In interference-free calibration data, only data meeting the requirements are used. Calculate the normalized small eigenvalues ​​of the effective samples Its 99th percentile is denoted as and take ,in It is acceptable When the system starts up but before a complete long time window is formed, the temporary noise power estimate is determined based on the short time window covariance matrix in the same way; after the long time window is formed, the system switches to long time window noise power estimation.

[0048] The decision threshold is obtained by obtaining large eigenvalues. Then, the short-window covariance matrix and the long-window covariance matrix were analyzed respectively. The effective rank of the short window and the effective rank of the long window are counted according to the following rules: as well as Calculation. Among them, and The first and second are the short-window covariance matrix and the long-window covariance matrix, respectively. One eigenvalue; The cardinality of a set is represented by the eigenvalue index that satisfies the corresponding threshold condition. The quantity.

[0049] The effective rank of the short window and the effective rank of the long window obtained in step 5 are also used as one of the observations for joint decision-making in the subsequent step 6.

[0050] See Figure 4 This paper illustrates the effective rank responses of the short-window and long-window covariance matrices under three interference scenarios: two fixed-polarization interference sources, a rotating variable-polarization single interference source, and a switching variable-polarization single interference source. The identifiable premise for both the short-window and long-window effective rank of two fixed-polarization interference sources to be 2 is that the array response vectors corresponding to the two interference sources are linearly independent, both interference components form eigenvalues ​​exceeding the large eigenvalue judgment threshold, and they are not completely coherent within the corresponding observation window. If the above premises are not met, and the measured effective rank does not reach 2, then the corresponding judgment correspondence is entered according to the measured effective rank; if a match cannot be found, it enters an unmatched state. Under the rotating variable-polarization single interference source condition, the long-window covers the polarization state change process, and the long-window effective rank is 2. The short-window effective rank is 1 or 2 depending on the magnitude of the polarization state change within the short-window. Under the switching polarization single interference source condition, the polarization state within a single short time window remains in a switching state, and the effective rank of the short time window is 1. The long time window covers multiple polarization switching states, and the effective rank of the long time window is 2.

[0051] Step 6: Determine the interference situation category corresponding to the received signal based on the rotation angular rate, rotation direction consistency index, effective rank of the short window, and effective rank of the long window.

[0052] The rotation angular rate obtained in step 3 The rotation direction consistency index obtained in step 4 and the effective rank of the short-time window obtained in step 5 and the effective rank of long time windows The combined observations are used to determine the interference situation category corresponding to the received signal according to the following four decision correspondences: The first type of judgment correspondence: In , And rotational angular rate When the received signal falls within the preset reference jitter range, the interference situation category is determined to be single fixed polarization interference. Because... As a non-negative quantity, the preset reference jitter range is determined to be... Continuous acquisition under interference-free conditions or static reference interference conditions. The rotation angular rate samples corresponding to each covariance update time, among which Let the allowable baseline jitter out-of-bounds probability be... , The sample The quantile is determined as the baseline jitter upper limit. .

[0053] The second type of judgment correspondence: In , , When the main characteristic direction remains stable, the interference situation category corresponding to the received signal is determined to be two fixed polarization interferences. The stability of the main characteristic direction is determined by the stationary deviation. Judgment, and Continuous acquisition under static reference interference conditions The principal eigenvectors of the unit norm will be obtained a static deviation The quantile is determined as the static judgment margin. ,in During operation, when The direction of the main feature is determined to remain stable. The static determination used for adjacent covariance update times, and the stability test within a short time window. Sample and calibrate separately.

[0054] The third type of judgment correspondence: according to Calculate the sliding window mean of the rotational angular rate. For 1 or 2 , In continuous All observation windows showed values ​​greater than the rotation threshold. And the consistency index of rotation direction In continuous All observation windows showed values ​​greater than the consistency decision threshold. At that time, the interference situation category corresponding to the received signal is determined to be a rotating variable polarization single interference source. The sliding window and the direction consistency observation window use the same... There are several update intervals. Under conditions of single fixed polarization interference, two fixed polarization interferences, and other non-rotating reference interferences, let the allowable single observation window rotation misjudgment probability be . ,in, , respectively and Sample The quantile is determined as follows: and Therefore and Next, count the number of consecutive observation windows that simultaneously exceed their respective thresholds. ,according to Determine the number of continuous confirmation windows for rotation features. . Number of delayed confirmation windows for the final category Once determined, a joint verification is performed according to the total response delay formula described in the following hysteresis decision paragraph.

[0055] The fourth type of judgment correspondence: setting a switch candidate hold counter. When the rotation direction consistency index The value is valid, and the third type of judgment relationship is not satisfied. , and season In other cases, let .exist It is a valid value, does not satisfy the third type of judgment correspondence and If the current single-window initial decision result is satisfied, set it to the switching-type variable polarization single interference source; if the third decision correspondence is satisfied, prioritize the use of the single-window initial decision result of the rotating-type variable polarization single interference source and clear it to zero. .in, The preset jump threshold is determined using calibration data under interference-free and static reference interference conditions: based on the calculated reference jump metric, the allowable reference condition jump misjudgment probability is set to... , The measurement samples under each benchmark condition The larger value among the quantiles is denoted as When the static criterion margin is satisfied When selecting the state separation protection quantity ,make and take Thus ensuring For example, Desirable .when Or there is no such range. If the current calibration data cannot form separable static thresholds and transition thresholds, the fourth decision correspondence remains mismatched, and the reference data is re-acquired for calibration.

[0056] The above four types of judgment correspondence are matched in the order of the third, fourth, second, and first types of judgment correspondence. When When the value is invalid, the third and fourth decision correspondences are not executed. If the rotation angular rate, effective rank of the short window, effective rank of the long window, and corresponding static deviation are valid, the first or second decision correspondence can still be matched. When the current effective joint decision observation does not satisfy any executable decision correspondence, the current single-window initial decision result is recorded as an unmatched state. The unmatched state is not considered a new interference situation category, the continuity consistency count of the current hysteresis decision is cleared, and the last determined final output category remains unchanged. When the system starts for the first time and has not yet formed a historical final output category, the output state is kept as a pending decision state, and no dynamic characteristic parameters are output.

[0057] To suppress output jitter caused by observation fluctuations, a hysteresis decision is applied to the initial decision result of the single window: when continuous When the initial decision results of each observation window are all of the same valid disturbance situation category, the final output category is updated to that category; when consecutive results are not completely identical or a mismatch occurs, the current consecutive consistency count is cleared and the previously determined final output category remains unchanged. This represents the number of delayed confirmation windows for the final category. In calibration data where the disturbance situation category is known, this is the maximum number of observation windows whose initial decision result is consecutively misclassified to the same erroneous category while the actual category remains unchanged. ,according to Sure .set up The update time interval between two consecutive initial decision results of a single window. The maximum allowable category response latency; the formation of the rotation direction consistency index requires... Each update interval, rotation feature continuously confirmed requires addition Each update interval, delayed confirmation requires addition Each update interval, therefore according to Verify the total response latency of the rotating final category, and according to The conservative response latency for the final category of the verification switch is determined. If any verification condition is not met, the latency is reduced. Or recalibrate after adjusting the corresponding decision threshold. and If the conditions are still not met after recalibration, maintain the pending state and do not update the final output category and dynamic feature parameters. Switch the candidate hold counter so that a valid transition triggers continuous... A switching candidate result is obtained, thereby enabling the hysteresis confirmation to be completed.

[0058] See Figure 5 It demonstrates the joint decision state transition and delay mechanism. Figure 5 The four state nodes correspond to a single fixed-polarization interference source, two fixed-polarization interference sources, a rotating variable-polarization single interference source, and a switching variable-polarization single interference source, respectively. The solid line transition arrows between state nodes indicate continuous... When the initial decision results of each observation window match the target category, the final output category transitions from the current state to the target state. The dashed guide arrows in the figure indicate that if the transition condition is not met or the current initial decision result of the single window is unmatched, the final output category remains unchanged. Therefore, Figure 5 The hysteresis mechanism shown can prevent the decision fluctuations of a single observation window from directly causing the final output category switch.

[0059] After determining the final output category through hysteresis decision, the corresponding dynamic feature parameter extraction method is selected based on the final output category. Let the dynamic feature observation window be... ,in, ,and When the total response delay verification formula is true, at least one value can be taken. When the final output category is a rotating variable polarization single interference source, according to... Calculate an estimate of the polarization change rate.

[0060] When the final output category is a switched polarization single interference source, Adjacent consecutive observation windows are merged into a single candidate switching event. In the observation sequence containing each candidate switching event, the following conditions will be met: Each set of the largest consecutive observation windows is denoted as a stable segment in chronological order. , and demand ,in, For the number of stable segments, , This indicates the number of covariance update times included in the stable segment; The principal eigenvector of the unit norm is determined as the representative direction of the stable segment. The state is established by representing the direction with the first stable segment. And recorded as To be the first to meet The stable segment represents the direction establishment state. And recorded as ; thereafter only the state or state The representative directional distance is not greater than A stable segment is uniquely assigned to its corresponding state. If any stable segment simultaneously satisfies the classification conditions for two states or fails to satisfy either condition, the current period estimate is recorded as invalid. Only when all stable segments can be uniquely classified and the state labels alternate in chronological order are candidate switching events located between two adjacent stable segments of different states determined as valid switching events. The moment corresponding to the maximum value is taken as the switching moment, and these moments are arranged in chronological order to form a valid switching moment sequence. At the same time Estimate polarization residence time. When At that time, according to Calculate the states of two consecutive states or two adjacent states The estimated complete handover cycle between them is used as the handover type. . To effectively switch the number of events, This indicates that the sample median is taken; the two-state alternation confirmation condition is not met or At this time, the switching cycle parameter is kept in an pending update state. Only the confirmed valid dynamic feature parameters and the final output category are output to the subsequent CRPA anti-interference processing chain as adaptive trigger parameters for window length, sample selection strategy or weight refresh cycle.

[0061] In summary, the satellite navigation anti-jamming array polarization interference detection and discrimination method provided in this embodiment can discriminate interference situation categories, including single fixed polarization interference, two fixed polarization interference, rotating variable polarization single interference source, and switching variable polarization single interference source, by processing the signal of the received data, and simultaneously output dynamic feature parameters for adaptive triggering of the subsequent CRPA anti-jamming processing chain.

[0062] Example 2: Corresponding to Example 1, this example provides a satellite navigation anti-jamming array polarization interference detection and discrimination device, including... Figure 6The array element receiving unit 100, dual-time-scale covariance estimation unit 200, principal feature analysis unit 300, rotation angular rate and consistency index calculation unit 400, effective rank calculation unit 500, joint decision unit 600, and decision result output interface 700 are shown. The dual-time-scale covariance estimation unit 200, principal feature analysis unit 300, rotation angular rate and consistency index calculation unit 400, effective rank calculation unit 500, and joint decision unit 600 are implemented by at least one of a field-programmable gate array, a digital signal processor, or a central processing unit. The output side of the processing platform is connected to the subsequent CRPA anti-interference processing unit 800 through the decision result output interface 700.

[0063] 1. Array element receiving unit 100 The array element receiving unit 100 includes a satellite navigation single-polarized antenna array. 101 single-polarization array elements, and with Each single-polarization array element 101 is connected one-to-one to the radio frequency front-end 102 and the down-conversion and sampling channel 103, wherein... It is a positive integer not less than 4. The array element receiving unit 100 is used for... The received signals of each single-polarization array element 101 are down-converted and sampled respectively, and the output of the down-converted and sampled channel 103 is... The digital samples of the digital baseband signal at the same sampling time are combined as follows: The complex-dimensional snapshot vector is generated by combining snapshot vectors corresponding to different sampling times in chronological order into a snapshot vector sequence, and then outputting the snapshot vector sequence to the dual-time-scale covariance estimation unit 200. Each single-polarization array element 101 has the same expected polarization response; the RF front-end 102 is disposed in A single-polarization array element 101 is located between the downconversion and sampling channel 103, and is used to perform radio frequency filtering and amplification.

[0064] 2. Dual-time-scale covariance estimation unit 200 The dual-timescale covariance estimation unit 200 is connected to the array element receiving unit 100 and is used to receive the snapshot vector sequence output by the array element receiving unit 100. It estimates the sample covariance of the snapshot vector sequence in parallel according to the short time window length and the long time window length, and outputs the short time window covariance matrix and the long time window covariance matrix. The long time window length is greater than the short time window length.

[0065] The dual-timescale covariance estimation unit 200 includes a parallel short-window covariance estimation subunit 201 and a long-window covariance estimation subunit 202. The short-window covariance estimation subunit 201 estimates the sample covariance of the snapshot vector sequence according to the short-window length and outputs a short-window covariance matrix. The long-window covariance estimation subunit 202 estimates the sample covariance of the snapshot vector sequence according to the long-window length and outputs a long-window covariance matrix. The short-window and long-window covariance estimation subunits 201 and 202 can estimate the sample covariance using a time-averaged method of snapshot vectors within a sliding window, or they can use a recursive update method with an exponential forgetting factor. When using the recursive update method, the corresponding exponential forgetting factor is determined based on the target equivalent sample size. The outputs of the short-window and long-window covariance estimation subunits 201 and 202 are respectively sent to the main feature analysis unit 300 via an internal bus.

[0066] 3. Main Feature Analysis Unit 300 The main feature analysis unit 300 is connected to the dual-time-scale covariance estimation unit 200. It is used to perform eigenvalue decomposition on the short-time window covariance matrix and the long-time window covariance matrix respectively, and output the main eigenvector corresponding to the largest eigenvalue in the short-time window covariance matrix and the two eigenvectors corresponding to the two largest eigenvalues ​​in the long-time window covariance matrix. Based on the two eigenvectors, a two-dimensional interference subspace is constructed, and the eigenvalue sets of the short-time window covariance matrix and the long-time window covariance matrix are output respectively.

[0067] The main feature analysis unit 300 includes a parallel short-window eigenvalue decomposition subunit 301 and a long-window eigenvalue decomposition subunit 302. The short-window eigenvalue decomposition subunit 301 performs eigenvalue decomposition on the short-window covariance matrix, outputting the principal eigenvector corresponding to the largest eigenvalue in the short-window covariance matrix and the eigenvalue set of the short-window covariance matrix. The long-window eigenvalue decomposition subunit 302 performs eigenvalue decomposition on the long-window covariance matrix, obtaining the eigenvalue set and corresponding eigenvector set of the long-window covariance matrix, and arranging the obtained eigenvalues ​​in descending order of value; it also extracts the two eigenvectors corresponding to the two largest eigenvalues ​​and constructs a two-dimensional interference subspace based on these two eigenvectors. The output of the main feature analysis unit 300 is sent to the rotation angular rate and consistency index calculation unit 400 and the effective rank calculation unit 500, respectively.

[0068] 4. Rotational angular rate and consistency index calculation unit 400 The rotation angular rate and consistency index calculation unit 400 is connected to the main feature analysis unit 300 and is used to calculate the rotation angular rate and rotation direction consistency index of the main feature vector based on the main feature vector and the two-dimensional interference subspace output by the main feature analysis unit 300.

[0069] The rotation angular rate and consistency index calculation unit 400 includes a rotation angular rate calculation subunit 401 and a rotation direction consistency index calculation subunit 402. The rotation angular rate calculation subunit 401 is used to normalize the principal eigenvectors corresponding to adjacent covariance update times using the unit norm, according to... Calculate the angle between adjacent principal feature directions, and follow the... The rotation angular rate is calculated; the rotation direction consistency index calculation subunit 402 is used to obtain a continuous alignment basis according to the global alignment method of non-degenerate correlation matching or near-degenerate two-dimensional basis matrix as described in Embodiment 1, project the principal eigenvector onto the two-dimensional interference subspace and calculate the projection angle sequence, take the sign of the surround correction increment of adjacent projection angles and calculate the absolute value of the sign mean to obtain the rotation direction consistency index. The output of the rotation angular rate and consistency index calculation unit 400 is sent to the joint decision unit 600.

[0070] 5. Effective Rank Calculation Unit 500 The effective rank calculation unit 500 is connected to the main feature analysis unit 300. Based on the eigenvalue sets of the short-window and long-window covariance matrices and the large eigenvalue decision threshold, it counts the number of eigenvalues ​​exceeding the large eigenvalue decision threshold in both matrices and outputs the effective rank of the short-window and long-window covariance matrices. The large eigenvalue decision threshold is determined by multiplying the noise power estimate and the threshold coefficient as described in Example 1.

[0071] The effective rank calculation unit 500 includes a short-window effective rank counting subunit 501 and a long-window effective rank counting subunit 502. The short-window effective rank counting subunit 501 counts the eigenvalues ​​of the eigenvalue set of the short-window covariance matrix according to the large eigenvalue decision threshold, and outputs the short-window effective rank. The long-window effective rank counting subunit 502 counts the eigenvalues ​​of the eigenvalue set of the long-window covariance matrix according to the large eigenvalue decision threshold, and outputs the long-window effective rank. The output of the effective rank calculation unit 500 is sent to the joint decision unit 600.

[0072] 6. Joint Judgment Unit 600 The joint decision unit 600 is connected to the rotation angular rate and consistency index calculation unit 400 and the effective rank calculation unit 500, respectively. It is used to determine the interference situation category corresponding to the received signal based on the rotation angular rate, rotation direction consistency index, short window effective rank and long window effective rank, and output the interference situation category and the dynamic characteristic parameters corresponding to the interference situation category.

[0073] The joint decision unit 600 includes a decision table matching subunit 601 and a hysteresis decision subunit 602. The decision table matching subunit 601 performs matching in the following order: rotating variable polarization single interference source, switching variable polarization single interference source, two fixed polarization interferences, and a single fixed polarization interference. Internally, it is equipped with a switching candidate hold counter, which maintains continuity after a valid principal characteristic vector jump trigger, as described in Embodiment 1. A switching candidate result is given, and when the rotating decision condition is met, the rotating candidate is output first and the holding is terminated. The hysteresis decision subunit 602 continuously... When all candidate results belong to the same valid category, update the final output category; when a candidate result changes or becomes unmatched, clear the current continuous consistency count and keep the last determined final output category unchanged; when the system starts for the first time and has not yet formed a historical final output category, keep it in a pending state.

[0074] The interference situation categories output by the joint decision unit 600 include single fixed polarization interference, two fixed polarization interference, rotating variable polarization single interference source, and switching variable polarization single interference source. The dynamic characteristic parameters output by the joint decision unit 600 include an estimated value of the polarization change rate when it is determined to be a rotating variable polarization single interference source, and an estimated value of the switching period when it is determined to be a switching variable polarization single interference source.

[0075] 7. Judgment result output interface 700 The decision result output interface 700 is connected to the joint decision unit 600, and is used to output the interference situation category and dynamic characteristic parameters output by the joint decision unit 600 to the subsequent CRPA anti-interference processing unit 800. The decision result output interface 700 and the subsequent CRPA anti-interference processing unit 800 are connected via a software interface. The interference situation category and dynamic characteristic parameters output by the decision result output interface 700 serve as adaptive trigger parameters for the subsequent CRPA anti-interference processing unit 800, driving the subsequent CRPA anti-interference processing unit 800 to adaptively adjust according to the current interference situation category in sample selection, weight refresh cycle, and zero-traps maintenance strategy.

[0076] The specific execution logic, data processing order, parameter meaning, and calculation method of each unit and its subunits in this embodiment can be found in the processing procedures corresponding to their functions in the method described in Embodiment 1. Specifically, the array element receiving unit 100 corresponds to step 1 in Embodiment 1; the dual-timescale covariance estimation unit 200 corresponds to step 2; the principal feature analysis unit 300 and the rotation angular rate calculation subunit 401 correspond to step 3; the long-time-window eigenvalue decomposition subunit 302 and the rotation direction consistency index calculation subunit 402 correspond to step 4; the effective rank calculation unit 500 corresponds to step 5; the joint decision unit 600 corresponds to the joint decision and hysteresis processing in step 6; and the decision result output interface 700 corresponds to the output process of the final output category and its corresponding dynamic feature parameters in Embodiment 1. The technical content in Example 1 regarding the estimation of short-window and long-window covariance matrices, extraction of principal eigenvectors and two-dimensional interference subspaces, calculation of rotation angular rate and rotation direction consistency index, determination of short-window and long-window effective rank, handling of mismatched states, hysteresis decision, and output of dynamic feature parameters are all applicable to this example, provided they do not contradict each other.

[0077] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting and discriminating variable polarization interference in satellite navigation anti-jamming arrays, characterized in that, include: right The received signals from each single-polarization array element are down-converted and sampled to obtain a snapshot vector sequence. It is an integer not less than 4; The sample covariance of the snapshot vector sequence is estimated in parallel using short and long time windows, resulting in the short-window covariance matrix and the long-window covariance matrix, with the long window length being greater than the short window length. Eigenvalue decomposition is performed on the two covariance matrices respectively, and the principal eigenvector of the unit norm corresponding to the largest eigenvalue of the short time window and the eigenvectors corresponding to the two largest eigenvalues ​​of the long time window are extracted. The long time window eigenvectors span a two-dimensional interference subspace. The rotation angular rate is calculated based on the angle between adjacent principal feature vectors and the time interval between the centers of short time windows; the principal feature vectors are projected onto a continuously aligned two-dimensional interference subspace, and the rotation direction consistency index is calculated based on the surround correction angle increment of the projection angle sequence. The threshold is determined by using the product of the noise power estimate and the threshold coefficient as the largest eigenvalue. The number of eigenvalues ​​exceeding the threshold in the two covariance matrices is counted to obtain the effective rank of the short window and the effective rank of the long window, respectively. The interference situation category is determined based on the rotation angular rate, rotation direction consistency index, and effective rank of short and long time windows, including single fixed polarization interference, two fixed polarization interference, rotating variable polarization single interference source, and switching variable polarization single interference source. When determined to be a rotating variable polarization single interference source, output the estimated polarization change rate; when determined to be a switching variable polarization single interference source, and the stable segments before and after the candidate switching event and between adjacent events can be classified into two discrete polarization states. , Furthermore, when the status labels alternate over time, the estimated switching cycle value is output, and the switching cycle parameter is kept in a pending update state; the interference situation category and the confirmed valid dynamic characteristic parameters are used as adaptive trigger parameters for the CRPA anti-interference processing chain.

2. The satellite navigation anti-interference array polarization interference detection and discrimination method according to claim 1, characterized in that, The methods for determining the type of interference situation include: When the effective rank of the short window is 1, the effective rank of the long window is 1, and the rotation angular rate falls within the preset reference jitter range, the interference situation is determined to be a single fixed polarization interference. When the effective rank of the short window is 2, the effective rank of the long window is 2, the rotational angular rate falls within the preset reference jitter range, and the main feature direction remains stable, the interference situation category is determined to be two fixed polarization interferences. When the effective rank of the short time window is 1 or 2, the effective rank of the long time window is 2, and the mean of the rotation angular rate is in continuous... Within each observation window, the rotation threshold was exceeded and the rotation direction consistency index was continuously... When the consistency decision threshold is exceeded within each observation window, the interference situation is determined to be a rotating variable polarization single interference source. The number of windows is continuously confirmed for rotation features, and ; The rotation direction consistency index is valid, the judgment condition for a single interference source with rotational polarization is not met, the effective rank of the short window is 1, the effective rank of the long window is 2, and the main characteristic vector jump metric is valid. Exceeding the preset transition threshold Time-triggered switching candidate retention; continuous since the trigger observation window. Within each observation window, if the rotation direction consistency index is valid and the criteria for determining a rotating variable polarization single interference source are not met, the initial decision result of the single window will remain as a switching variable polarization single interference source; among which, , and These are the principal eigenvectors of the unit norm corresponding to adjacent short time windows. The final category of delayed confirmation window, and .

3. The satellite navigation anti-interference array polarization interference detection and discrimination method according to claim 2, characterized in that, The method for determining the type of interference situation also includes hysteresis decision; hysteresis decision includes: when continuous When the initial decision results of each observation window are all of the same disturbance situation category, the final output category will be updated to the disturbance situation category; when consecutive When the initial decision results of each observation window are not completely identical, the previously determined final output category remains unchanged; when an unmatched state occurs, the current continuous consistency count is cleared and the previously determined final output category remains unchanged; when the system starts for the first time and has not yet formed a historical final output category, it remains in a pending decision state; among these, The final category of delayed confirmation window, and .

4. The satellite navigation anti-interference array polarization interference detection and discrimination method according to claim 1, characterized in that, The calculation method for the rotation direction consistency index includes: taking the sign of the angle increments at adjacent times in the projection angle sequence to obtain a sign sequence, and using the absolute value of the mean of the sign sequence within the observation window as the rotation direction consistency index.

5. The satellite navigation anti-interference array polarization interference detection and discrimination method according to claim 1, characterized in that, Short-term window length is recorded as Long-term window duration is recorded as ,and Short-term window length meets ,in, The dimensionless sampling redundancy coefficient is between 5 and 10; the short-time window length is less than the number of samples corresponding to the expected polarization dwell time or the expected polarization change period; the sampling interval between adjacent snapshots is set to... The expected polarization change period is The preset period coverage is Long-term window length satisfies ,in For a single interference source with switching polarization, the two different polarization states are denoted as states. and state Long-term window at least covers the state ,state and two adjacent states or two adjacent states The complete switching cycle between them.

6. The satellite navigation anti-interference array polarization interference detection and discrimination method according to claim 1, characterized in that, Methods for determining noise power estimation include: sorting the eigenvalue set of the long-time window covariance matrix in descending order of value, and estimating the effective signal subspace dimension using the minimum description length criterion or the Akaike information content criterion. The eigenvalues ​​that are not included in the effective signal subspace after sorting are determined as the small eigenvalue set. and will The mean of each characteristic value is used as the noise power estimate; when the noise power estimate is not greater than a preset positive number... When the current observation window is not yet matched, the temporary noise power estimate is determined based on the short-window covariance matrix in the same way during the system startup phase.

7. The satellite navigation anti-interference array polarization interference detection and discrimination method according to claim 3, characterized in that, This also includes calibrating the thresholds and the number of confirmation windows; collecting the distribution of rotation angular rate and rotation direction consistency index under interference-free or static reference interference conditions, determining the rotation threshold based on the preset quantiles of the rotation angular rate samples, determining the consistency decision threshold based on the preset quantiles of the rotation direction consistency index samples, and determining the threshold coefficient based on the statistical distribution of the small eigenvalue set; and, under non-rotational reference interference conditions, counting the maximum number of consecutive observation windows where both the rotation angular rate exceeds the rotation threshold and the rotation direction consistency index exceeds the consistency decision threshold. ,according to Determine the number of continuous confirmation windows for rotational features; count the maximum number of observation windows whose initial decision results for a single window are consecutively misclassified into the same error category. ,according to Determine the final category hysteresis confirmation window number; assume the observation window for the rotation direction consistency index includes The time interval between two consecutive single-window initial decision results is [number] projection angles. The maximum allowed category response latency is ,according to Verify the total response latency of the rotating final category, and according to Conservative response latency for the final category of the verification switch.

8. The satellite navigation anti-interference array polarization interference detection and discrimination method according to claim 2, characterized in that, The output methods for dynamic feature parameters include: when the determined interference situation category is a rotating variable polarization single interference source, the average rotation angular rate within the dynamic feature observation window is output as an estimate of the polarization change rate; when the determined interference situation category is a switching variable polarization single interference source, adjacent observation windows where the main feature vector jump metric continuously exceeds a preset jump threshold are merged into a candidate switching event; in the observation sequence containing each candidate switching event, the main feature vector jump metric is not greater than the static determination margin. Each set of the largest consecutive observation windows is denoted as a stable segment in chronological order. , and demand ,in, For the number of stable segments, , For short-term window length, The sampling interval between adjacent snapshots. The short time window center time interval between adjacent covariance update times; The principal eigenvector of the unit norm is determined as the representative direction of the stable segment. The state is established by representing the direction with the first stable segment. Record it as To be the first to meet The stable segment represents the direction establishment state. Record it as ,in, ; thereafter only the state or state The representative direction distance or Not greater than A stable segment is uniquely assigned to its corresponding state. Only when there is no stable segment that cannot be uniquely classified and the state labels alternate in chronological order, a candidate switching event located between two adjacent stable segments of different states is determined as a valid switching event. The moment corresponding to the maximum value of the main feature vector jump metric within this event is taken as the switching moment, and a sequence of valid switching moments is formed in chronological order. ;when At that time, according to Determine and output the estimated value of the complete switching cycle; otherwise, leave the switching cycle parameter as pending update.

9. A satellite navigation anti-interference array polarization interference detection and discrimination device, characterized in that, include: Array element receiving unit (100), used for receiving The received signals from each single-polarization array element (101) are down-converted and sampled to output a snapshot vector sequence. It is an integer not less than 4; The dual-time-scale covariance estimation unit (200) is used to estimate the sample covariance of the snapshot vector sequence in parallel using short and long time windows, and outputs the short and long time window covariance matrices, with the long time window being longer than the short time window. The main feature analysis unit (300) is used to perform eigenvalue decomposition on the two covariance matrices and output the unit norm main eigenvector corresponding to the maximum eigenvalue of the short time window, the two-dimensional interference subspace spanned by the eigenvectors corresponding to the two maximum eigenvalues ​​of the long time window, and the eigenvalue set of the two matrices. The rotation angular rate and consistency index calculation unit (400) is used to calculate the rotation angular rate based on the angle between adjacent principal feature vectors and the short time window center time interval, and to project the principal feature vectors onto the continuously aligned two-dimensional interference subspace, and to calculate the rotation direction consistency index based on the surround correction angle increment of the projection angle sequence. The effective rank calculation unit (500) is used to count the number of eigenvalues ​​in two matrices that exceed the threshold by using the product of the noise power estimate and the threshold coefficient as the threshold, and output the effective rank of the short window and the effective rank of the long window. The joint decision unit (600) is used to determine the interference situation category based on the rotation angular rate, rotation direction consistency index, short window effective rank and long window effective rank, and output the category and effective dynamic characteristic parameters. The categories include single fixed polarization interference, two fixed polarization interference, rotating variable polarization single interference source and switching variable polarization single interference source. The decision result output interface (700) is used to output the interference situation category and dynamic characteristic parameters to the subsequent CRPA anti-interference processing unit (800).

10. The satellite navigation anti-interference array polarization interference detection and discrimination device according to claim 9, characterized in that, The dual-timescale covariance estimation unit (200), the principal feature analysis unit (300), the rotation angular rate and consistency index calculation unit (400), the effective rank calculation unit (500), and the joint decision unit (600) are implemented by at least one of a field-programmable gate array, a digital signal processor, or a central processing unit.

Images