CN106443628B - A kind of STAP method based on the estimation of concentration matrix nonlinear shrinkage - Google Patents
A kind of STAP method based on the estimation of concentration matrix nonlinear shrinkage Download PDFInfo
- Publication number
- CN106443628B CN106443628B CN201610857718.1A CN201610857718A CN106443628B CN 106443628 B CN106443628 B CN 106443628B CN 201610857718 A CN201610857718 A CN 201610857718A CN 106443628 B CN106443628 B CN 106443628B
- Authority
- CN
- China
- Prior art keywords
- matrix
- sample
- covariance matrix
- following formula
- estimation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
- G01S7/414—Discriminating targets with respect to background clutter
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a kind of STAP methods based on the estimation of concentration matrix nonlinear shrinkage comprising following steps: step 1: calculating sample covariance matrix;Step 2: carrying out feature decomposition to sample covariance matrix, characteristic value and feature vector are obtained, if sample number is greater than filter dimension and goes to third step, skips the 4th step;If sample number is less than filter dimension, third step is skipped, the 4th step is gone to;Step 3: calculating the corresponding shrinkage value of each sample covariance matrix characteristic value, and combine the feature vector of sample covariance matrix, estimated accuracy matrix;Step 4: calculating separately its shrinkage value, and combine the feature vector of sample covariance matrix, estimated accuracy matrix when sample covariance matrix characteristic value is greater than 0 and is equal to 0 two kinds;Step 5: calculating STAP filter factor, and handle reception data according to the concentration matrix of estimation.
Description
Technical field
The invention belongs in Radar Signal Processing clutter recognition and target detection technique field at a slow speed, more particularly to it is a kind of
Based on concentration matrix nonlinear shrinkage estimation space-time adaptive processing (Space Time Adaptive Processing,
STAP) method.
Background technique
Space-time adaptive, which handles (Space Time Adaptive Processing, STAP) technology, to be pressed down in airborne radar
One of the important method of clutter processed.STAP technology is usually inverted (Sample Matrix with sample covariance matrix
Inversion, SMI) method estimate filter factor.When using the above method, damaged to desired Signal Interference and Noise Ratio
It loses and is less than 3dB, then required independent same distribution sample number is twice of filter dimension.Due to time freedom degree and spatial degrees of freedom
Product it is larger, the estimation of miscellaneous covariance matrix of making an uproar needs a large amount of aid sample, and in practical application, aid sample number is often not
It is able to satisfy the demand of covariance matrix.
In the case where lacking more information, diagonal loading method is commonly used for solving the above problems.But common pair
Angle loading method load factor is frequently with empirical value, Jian Li et al. [L.Jian, D.Lin, and P.Stoica, " Fully
automatic computation of diagonal loading levels for robust adaptive
beamforming,"in Acoustics,Speech and Signal Processing,2008.ICASSP 2008.IEEE
International Conference on, 2008, pp.2325-2328] from the smallest angle of estimation error of the covarianee matrix
Load factor is solved, the quantum chemical method mode of load factor is given.But the diagonal loading method of Jian Li et al. is from association
The smallest angle of variance matrix evaluated error considers a problem, not from the concentration matrix (association closer with filtering weight coefficient relationship
Variance matrix it is inverse) from the point of view of problem;Also, in covariance matrix, to all spies of sample covariance matrix
Value indicative is handled different sample characteristics using different parameters by the way of unified parameters processing, is just had
Evaluated error may be further decreased.This makes such method have the space for further increasing performance.
Summary of the invention
The present invention In view of the above shortcomings of the prior art, propose it is a kind of based on concentration matrix nonlinear shrinkage estimation
STAP method, no matter sample number is more than or less than filter dimension, and this method all has preferable performance.
Technical scheme is as follows:
A kind of STAP method based on the estimation of concentration matrix nonlinear shrinkage, comprising the following steps:
Step 1: calculating p × p ties up sample covariance matrix;
Step 2: carrying out feature decomposition to sample covariance matrix, the characteristic value { (λ of its descending arrangement is obtained1,…,λp)}
And corresponding feature vector { (u1,…,up)}.Third step is gone to if sample number is greater than filter dimension, skips the 4th step;If
Sample number is less than filter dimension, then skips third step, go to the 4th step.
Step 3: calculating sample covariance matrix ith feature value λ according to the following formulaiCorresponding shrinkage value
Wherein, Re [], which refers to, takes real part to operate, cp=p/N, N are sample number, and
Wherein, σ2Refer to noise power.In turn, it if the feature vector of sample covariance matrix is arranged as matrix U, is obtained by following formula
To the concentration matrix P of estimationNL
PNL=UANLUH
Wherein,Diag expression diagonal matrix, H expression vector or matrix are total to
Yoke transposition.
Step 4: sample covariance matrix ith feature value λiWhen > 0, respective sample covariance matrix is calculated according to the following formula
The shrinkage value of characteristic value
Wherein, cp=p/N, N are sample number, and
Work as λiWhen=0, calculate according to the following formula
The concentration matrix P of estimation is calculated by following formula againNL
PNL=UANLUH
Wherein,Diag expression diagonal matrix, H expression vector or matrix are total to
Yoke transposition.
Step 5: obtaining concentration matrix P according to estimationNLAfterwards, STAP filter factor is calculated by following formula,
Wherein, a is steering vector, and H indicates the conjugate transposition of vector or matrix.
Later, data are received using above-mentioned filter process, after filtering, Signal Interference and Noise Ratio is
Wherein, s is the steering vector of desired signal, and has sHS=1.
Further, in the third step, sample covariance matrix ith feature value λiShrinkage formula beWhereinCalculation formula be
Further, in the 4th step, in sample covariance matrix eigenvalue λiIn the case of 0, characteristic value is shunk public
Formula isWhereinCalculation formula beIn sample covariance matrix feature
Value λiIn the case where 0,Calculation formula be
The beneficial effect of the method for the present invention, compared with prior art, no matter sample number is less than or greater than filter dimension, this
Inventive method all has preferable performance, improves the detectability to target at a slow speed.Also, compared with the existing technology, this hair
Bright method still has preferable performance under the conditions of higher miscellaneous noise ratio (such as under the conditions of land clutter).
Detailed description of the invention
Fig. 1 is the flow diagram of the method for the present invention.
Fig. 2 is the Signal Interference and Noise Ratio of inventive algorithm, diagonal loading method and optimal processing under different sample conditions
(SINR)。
The SINR of Fig. 3 is sample number when being 100 inventive algorithm, diagonal loading method and optimal processing;
Wherein (a) is the SINR of STAP output
It (b) is Fig. 3 (a) close to the enlarged drawing of clutter recess.
Fig. 4 is the SINR of inventive algorithm, diagonal loading method and optimal processing under the conditions of different CNR.
Specific embodiment
With reference to the accompanying drawing, the present invention is described in more detail.
It is of the present invention it is a kind of based on concentration matrix nonlinear shrinkage estimation STAP method, as shown in Figure 1, include with
Lower step:
The first step calculates sample covariance matrix S.
If reception data dimension is p, aid sample number is N, and aid sample is arranged as p × N-dimensional matrix
X=[x1,x2,…,xN]
Then sample covariance matrix is writeable are as follows:
S=XXH/N
Second step carries out feature decomposition to sample covariance matrix S, obtains the characteristic value { (λ of its descending arrangement1,…,
λp) and corresponding feature vector { (u1,…,up)}.Later, third step is gone to if sample number is greater than filter dimension, skipped
4th step;If sample number is less than filter dimension, third step is skipped, the 4th step is gone to.
Third step calculates according to the following formula if sample number N is greater than matrix dimension p
Wherein, cp=p/N, σ2For noise power.Later, sample covariance matrix ith feature value is calculated according to the following formula
Shrinkage value
In turn, if the feature vector of sample covariance matrix is arranged as matrix U, then the concentration matrix estimated by following formula
PNL
PNL=UANLUH
Wherein,Diag expression diagonal matrix, H expression vector or matrix are total to
Yoke transposition.
Step 4: working as λ if sample number N is less than matrix dimension piWhen > 0, calculate according to the following formula
Later, the shrinkage value of respective sample covariance matrix characteristic value is calculated according to the following formula
Work as λiWhen=0, calculate according to the following formula
Finally, calculating the concentration matrix P of estimation by following formulaNL
PNL=UANLUH
Wherein,Diag expression diagonal matrix, H expression vector or matrix are total to
Yoke transposition.
Step 5: estimation obtains concentration matrix PNLAfterwards, STAP filter factor is calculated by following formula
Wherein, a is steering vector, and H indicates the conjugate transposition of vector or matrix.Later, it is connect using above-mentioned filter process
Receive data.After filtering, Signal Interference and Noise Ratio is
Wherein, s is the steering vector of desired signal, and has sHS=1.
The implementation process of the method for the invention is further illustrated below by emulation experiment.In experiment, umber of pulse 16,
Array number is 16, and the dimension of concentration matrix is 256 × 256.The side view if radar is positive.
Fig. 2 is miscellaneous noise ratio (CNR) when being 50dB, on the direction 20m/s of clutter recess (see Fig. 3) different sample conditions
Under SINR.It can be seen that generally, the SINR of the method for the invention is greater than diagonal load, closer to optimal processing
As a result.Also, compared with diagonal loading method, with increasing for sample number, the better astringency of the method for the invention, performance
It is restrained to optimal processing rapidly.Even moreover, under compared with small sample said conditions (10~20), the performance of the method for the invention
Also due to diagonally loading 2dB~5dB;Later, increasing with sample number, diagonal loading method and the method for the invention performance
Gap it is increasing.Wherein sample number be 30~70 when, the method for the invention better than diagonal Loading Method about 13.7dB~
17dB;When being greater than filter dimension to sample number, the gap of diagonal loading method and the method for the invention performance reduces, still
When sample number is 400, the method for the invention performance is still better than diagonal loading method about 1.82dB.Thus, either in sample
When this number is less than or greater than filter dimension, the method for the invention performance is superior to diagonal loading method.
Fig. 3 is that sample number is 100, when CNR is 50dB, and each Doppler frequency (having been converted into respective objects speed) is corresponding
SINR.It can see from figure (a), either in clutter indent, or far from clutter indent, the method for the invention
Performance all closer to optimal processing.Scheming (b) is the enlarged drawing close to clutter recess, it can be seen that with diagonal loading method
It compares, the clutter recess of the method for the invention is narrower, closer to optimal processing.Also, close to clutter indent, the present invention
The method is better than diagonal loading method about 10dB.
Fig. 4 is sample number when being 50, on the direction 20m/s of clutter recess, the SINR under the conditions of different CNR.By Fig. 4
It is found that in CNR 15dB, the SINR of the method for the invention is only than diagonally loading high 0.08dB, but increasing with CNR,
The performance gap of diagonal loading method and the method for the invention is increasing.CNR be 70dB when, the method for the present invention better than pair
Angle Loading Method about 35dB.
The present invention is explained in detail above, it is clear that if essentially without be detached from inventive point of the invention and
Effect, obvious variations to those skilled in the art, are also all included in the scope of protection of the present invention.
Claims (1)
1. a kind of STAP method based on the estimation of concentration matrix nonlinear shrinkage, which comprises the following steps:
Step 1: calculating p × p ties up sample covariance matrix;
Step 2: carrying out feature decomposition to sample covariance matrix, the characteristic value { (λ of its descending arrangement is obtained1,…,λp) and phase
The feature vector { (u answered1,…,up)};Third step is gone to if sample number is greater than filter dimension, skips the 4th step;If sample
Number is less than filter dimension, then skips third step, go to the 4th step;
Step 3: calculating sample covariance matrix ith feature value λ according to the following formulaiCorresponding shrinkage value
Wherein, Re [], which refers to, takes real part to operate, cp=p/N, N are sample number, and
Wherein, σ2Refer to that noise power if the feature vector of sample covariance matrix is arranged as matrix U, is estimated in turn by following formula
The concentration matrix P of meterNL
PNL=UANLUH
Wherein,Diag indicates that diagonal matrix, H indicate that the conjugation of vector or matrix turns
It sets;
Step 4: sample covariance matrix ith feature value λiWhen > 0, respective sample covariance matrix feature is calculated according to the following formula
The shrinkage value of value
Wherein, cp=p/N, N are sample number, and
Work as λiWhen=0, calculate according to the following formula
The concentration matrix P of estimation is calculated by following formula againNL,PNL=UANLUH
Wherein,Diag indicates that diagonal matrix, H indicate that the conjugation of vector or matrix turns
It sets;
Step 5: obtaining concentration matrix P according to estimationNLAfterwards, the filter factor of STAP is calculated by following formula,
Wherein, a is steering vector, and H indicates the conjugate transposition of vector or matrix,
Later, data are received using above-mentioned filter process, after filtering, Signal Interference and Noise Ratio is
Wherein, s is the steering vector of desired signal, and has sHS=1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610857718.1A CN106443628B (en) | 2016-09-27 | 2016-09-27 | A kind of STAP method based on the estimation of concentration matrix nonlinear shrinkage |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610857718.1A CN106443628B (en) | 2016-09-27 | 2016-09-27 | A kind of STAP method based on the estimation of concentration matrix nonlinear shrinkage |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106443628A CN106443628A (en) | 2017-02-22 |
CN106443628B true CN106443628B (en) | 2019-03-05 |
Family
ID=58169463
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610857718.1A Active CN106443628B (en) | 2016-09-27 | 2016-09-27 | A kind of STAP method based on the estimation of concentration matrix nonlinear shrinkage |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106443628B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108008262B (en) * | 2017-11-29 | 2020-07-07 | 国网上海市电力公司 | Rapid partial discharge detection method |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105445703A (en) * | 2015-11-27 | 2016-03-30 | 西安电子科技大学 | Two-stage time space adaptive processing method for airborne radar time space echo data |
CN105929371A (en) * | 2016-04-22 | 2016-09-07 | 西安电子科技大学 | Airborne radar clutter suppression method based on covariance matrix estimation |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7038618B2 (en) * | 2004-04-26 | 2006-05-02 | Budic Robert D | Method and apparatus for performing bistatic radar functions |
-
2016
- 2016-09-27 CN CN201610857718.1A patent/CN106443628B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105445703A (en) * | 2015-11-27 | 2016-03-30 | 西安电子科技大学 | Two-stage time space adaptive processing method for airborne radar time space echo data |
CN105929371A (en) * | 2016-04-22 | 2016-09-07 | 西安电子科技大学 | Airborne radar clutter suppression method based on covariance matrix estimation |
Non-Patent Citations (4)
Title |
---|
Fully Automatic Computation of Diagonal Loading Levels for Robust Adaptive Beamforming;LIN DU等;《IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS》;20100131;第46卷(第1期);全文 |
基于STAP杂波抑制的天基雷达PRF设计;张乔等;《雷达科学与技术》;20111031;第9卷(第5期);全文 |
基于对角加载的STAP性能改善;刘聪锋等;《电子与信息学报》;20080430;第30卷(第4期);全文 |
空时自适应杂波抑制;伍勇;《中国博士学位论文全文数据库 信息科技辑》;20090915;全文 |
Also Published As
Publication number | Publication date |
---|---|
CN106443628A (en) | 2017-02-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106646344B (en) | A kind of Wave arrival direction estimating method using relatively prime battle array | |
CN102830387B (en) | Data preprocessing based covariance matrix orthogonalization wave-beam forming method | |
CN103984676A (en) | Rectangular projection adaptive beamforming method based on covariance matrix reconstruction | |
CN103837861B (en) | The Subarray linear restriction Adaptive beamformer method of feature based subspace | |
CN105319545B (en) | Improve the MIMO-OFDM radar waveform design methods of STAP detection performances | |
CN105204006A (en) | Beam forming method based on subspace interference-plus-noise covariance matrix reconstruction | |
KR101768587B1 (en) | Covariance matrix estimation method for reducing nonstationary clutter and heterogeneity clutter | |
CN109639333B (en) | Beam forming method based on effective reconstruction covariance matrix | |
CN108037494B (en) | Radar target parameter estimation method under impulse noise environment | |
CN102944870A (en) | Robust covariance matrix diagonal loaded adaptive beam-forming method | |
CN107462877B (en) | Sky wave radar sea clutter suppression method based on priori knowledge | |
CN109743078B (en) | Unmanned aerial vehicle frequency hopping remote control signal detection and reception method and device based on array antenna | |
CN103293517B (en) | Diagonal-loading robust adaptive radar beam forming method based on ridge parameter estimation | |
CN104811867B (en) | Microphone array airspace filter method based on array virtual extended | |
CN109245814A (en) | Adaptive beamformer method based on maximum likelihood resampling | |
CN105913044A (en) | Multiple signal classification method based on Sigmoid covariance matrix | |
CN111337873A (en) | DOA estimation method based on sparse array | |
CN106443628B (en) | A kind of STAP method based on the estimation of concentration matrix nonlinear shrinkage | |
CN110865345B (en) | Rapid self-adaptive pulse compression method | |
CN115022137A (en) | Artificial noise elimination method based on sub-component analysis | |
CN115372925A (en) | Array robust adaptive beam forming method based on deep learning | |
CN107167776A (en) | The adaptive beam-forming algorithm compensated based on subspace | |
CN100557630C (en) | Be applicable to the particle method of estimation of non-linear system status | |
CN106685507A (en) | Beam forming method based on Constrained Kalman in colored noise environment | |
CN110361697B (en) | Robust beam forming method based on covariance matrix hybrid reconstruction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |