CN112946564A - DOA estimation method and device of beam space based on DNN and computer storage medium - Google Patents

DOA estimation method and device of beam space based on DNN and computer storage medium Download PDF

Info

Publication number
CN112946564A
CN112946564A CN202110390921.3A CN202110390921A CN112946564A CN 112946564 A CN112946564 A CN 112946564A CN 202110390921 A CN202110390921 A CN 202110390921A CN 112946564 A CN112946564 A CN 112946564A
Authority
CN
China
Prior art keywords
array
dnn
covariance matrix
beam space
signal
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.)
Granted
Application number
CN202110390921.3A
Other languages
Chinese (zh)
Other versions
CN112946564B (en
Inventor
文才
吉原杰
彭进业
李展
乐明楠
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Northwestern University
Original Assignee
Northwestern University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Northwestern University filed Critical Northwestern University
Priority to CN202110390921.3A priority Critical patent/CN112946564B/en
Publication of CN112946564A publication Critical patent/CN112946564A/en
Application granted granted Critical
Publication of CN112946564B publication Critical patent/CN112946564B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G01S3/00—Direction-finders for determining the direction from which infrasonic, sonic, ultrasonic, or electromagnetic waves, or particle emission, not having a directional significance, are being received
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Remote Sensing (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)

Abstract

The embodiment of the invention discloses a DOA estimation method and a device of a beam space based on DNN and a computer storage medium; the method can comprise the following steps: acquiring a covariance matrix of array received signals; generating received signal data in a beam space according to the covariance matrix of the array received signals and the beam pointing covariance matrix of each angle interval in the angle space; inputting the received signal data in the beam space into a Deep Neural Network (DNN) which has completed training by utilizing a training data set; and obtaining a DOA estimated value by linear amplitude interpolation of the signal data output by the DNN.

Description

DOA estimation method and device of beam space based on DNN and computer storage medium
Technical Field
The embodiment Of the invention relates to the technical field Of signal processing, in particular to a method and a device for estimating Direction Of Arrival (DOA) Of a beam space based on Deep Neural Networks (DNN), and a computer storage medium.
Background
DOA estimation is an important research direction for array signal processing, and has been widely applied to various military and civil systems including wireless communication, astronomical observation, radar and sonar. DOA estimation is always developed towards the trend of improving precision and super-resolution, and the generalization capability of the DOA estimation to various unknown scenes such as array errors, low signal-to-noise ratio, limited snapshots and the like is enhanced. For the current conventional DOA estimation schemes, they are parametric schemes, that is, they all implement forward mapping from signal direction to array output under the assumption that the mapping is invertible. Based on the above assumptions, the array outputs are matched by a pre-formed mapping, thereby enabling direction estimation. The performance of such parameterization schemes depends to a large extent on the consistency between the two mappings, i.e. the forward mapping from the signal direction to the array output and the backward mapping from the array output to the signal direction during data acquisition.
However, various defects may exist in the array system, such as non-ideal sensor design, installation of the array, mutual interference between sensors, and influence by environmental factors. Therefore, in the actual signal estimation, the above defects all have great influence on the DOA estimation performance, and further cause the reduction of the estimation accuracy.
Although there are many existing schemes to describe the influence of various defects by simplifying the model, and propose the corresponding automatic calibration process to improve the accuracy of DOA estimation; however, the above simplified models for array errors are mathematically triggered with various additional assumptions, which deviate from the actual situation to different degrees and do not accurately address array defects.
Disclosure of Invention
In view of this, embodiments of the present invention are intended to provide a method, an apparatus, and a computer storage medium for estimating DOA in a DNN-based beam space; the adaptive capacity of the deep neural network to array errors is effectively utilized, the advantage of a beam space is utilized, the adaptive capacity to the array errors is further improved while the calculated amount is reduced, and therefore the DOA estimation precision is better improved.
The technical scheme of the embodiment of the invention is realized as follows:
in a first aspect, an embodiment of the present invention provides a method for estimating DOA in a DNN-based beam space, including:
acquiring a covariance matrix of array received signals;
generating received signal data in a beam space according to the covariance matrix of the array received signals and the beam pointing covariance matrix of each angle interval in the angle space;
inputting the received signal data in the beam space into a Deep Neural Network (DNN) which has completed training by utilizing a training data set;
and obtaining a DOA estimated value by linear amplitude interpolation of the signal data output by the DNN.
In a second aspect, an embodiment of the present invention provides an apparatus for estimating DOA in a DNN-based beam space, where the apparatus includes: the method comprises the steps of obtaining a part, a generating part, a deep neural network DNN and an interpolation part; wherein the acquisition section is configured to acquire a covariance matrix of array reception signals;
the generation part is configured to generate the received signal data in the beam space according to the covariance matrix of the array received signals and the beam pointing covariance matrix of each angle interval in the angle space;
the DNN configured to input and output received signal data within the beam space;
the interpolation part is configured to obtain the DOA estimated value by linear amplitude interpolation of the signal data output by the DNN.
In a third aspect, an embodiment of the present invention provides a computing device, including a receiving array, a memory, and a processor; wherein the content of the first and second substances,
the receiving array is used for receiving an incident far-field signal;
the memory for storing a computer program operable on the processor;
the processor, when running the computer program, is configured to perform the steps of the method for DOA estimation based on DNN beam space of the first aspect.
In a fourth aspect, an embodiment of the present invention provides a computer storage medium storing a DNN-based beam space DOA estimation program, which when executed by at least one processor implements the steps of the DNN-based beam space DOA estimation method of the first aspect.
The embodiment of the invention provides a DOA estimation method, a device and a computer storage medium of a beam space based on DNN; the received signals are converted to the beam space from the array element space, so that the generalization capability of array errors is improved while the calculated amount is reduced and the resolution threshold of the signal-to-noise ratio is reduced; in addition, the DOA estimation is carried out by adopting the deep neural network, so that the generalization capability of the technical scheme to array errors can be further improved; moreover, because the technical scheme carries out spectrum estimation in a virtual beam space, even if array errors are not considered in the DNN training process, the method still has strong adaptability to the errors.
Drawings
Fig. 1 is a schematic flowchart of a DOA estimation method based on DNN beam space according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a DNN hierarchy provided by an embodiment of the present invention;
FIG. 3 is a graph illustrating angle estimation bias comparison results under different SNR conditions according to an embodiment of the present invention;
FIG. 4 is a graph of angle estimates and estimation error comparisons at an angle interval of 9.4 provided by an embodiment of the present invention;
FIG. 5 is a graph of angle estimates and estimation error versus angle intervals of 16.4 and 60 provided by an embodiment of the present invention;
FIG. 6 is a graph illustrating angle estimates in single-signal and three-signal scenarios according to an embodiment of the present invention;
FIG. 7 is a graph of the RMSE comparison of direction of arrival estimates for two signals from 31.5 and 41.5 orientations in the presence of different array defects as provided by an embodiment of the present invention;
fig. 8 is a schematic composition diagram of a DNN-based DOA estimation apparatus for beam space according to an embodiment of the present invention;
fig. 9 is a schematic composition diagram of another DNN-based beam space DOA estimation apparatus according to an embodiment of the present invention;
fig. 10 is a schematic hardware composition diagram of a computing device according to an embodiment of the present invention.
Detailed Description
The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.
It should be noted that, in order to solve the problem of the conventional parameterization scheme that depends on consistency of two mappings, i.e. input and output, and to improve generalization capability for array errors while reducing the amount of computation and lowering the threshold of signal-to-noise ratio resolution, refer to fig. 1, which shows a method for estimating DOA in a DNN-based beam space according to an embodiment of the present invention, where the method may include:
s101: acquiring a covariance matrix of array received signals;
s102: generating received signal data in a beam space according to the covariance matrix of the array received signals and the beam pointing covariance matrix of each angle interval in the angle space;
s103: inputting the received signal data in the beam space into a Deep Neural Network (DNN) which has completed training by utilizing a training data set;
s104: and obtaining a DOA estimated value by linear amplitude interpolation of the signal data output by the DNN.
Through the technical scheme shown in fig. 1, the received signals are converted from the array element space to the beam space, so that the generalization capability of array errors is improved while the calculated amount is reduced and the resolution threshold of the signal-to-noise ratio is reduced; in addition, the DOA estimation is carried out by adopting the deep neural network, so that the generalization capability of the technical scheme to array errors can be further improved; moreover, because the technical scheme carries out spectrum estimation in a virtual beam space, even if array errors are not considered in the DNN training process, the method still has strong adaptability to the errors.
For the technical solution shown in fig. 1, in some possible implementations, the obtaining a covariance matrix of array received signals includes:
based on the incidence of K independent far-field signals to a uniform array with M array elements, the incidence direction of the kth far-field signal is set to be thetakThe k far field signal received by the uniform array is sk(t)；
Sampling the received far-field signal by N sampling moments with the same interval to obtain a snapshot signal X ═ X (t) of the array receiving signal1),...,x(tN)]；
Wherein the content of the first and second substances,
Figure BDA0003016718860000051
v (t) is zero-mean white gaussian noise, and a (θ) represents an ideal steering vector;
calculating a covariance matrix of the array received signals according to:
Rxx＝E[x(tN)xH(tN)]＝ASAH+RN
wherein, E [. C]And (·)HRespectively representing a desired operator and a conjugate transform operator; a is an array steering matrix, and A ═ a (θ)1),a(θ2),...,a(θk)](ii) a S and RNThe covariance matrix and the noise matrix, respectively, of the ideal received signal are defined as: s ═ E [ S (t) SH(t)]，RN＝E[v(t)vH(t)]R since the noise follows a zero mean Gaussian distributionN＝σ2I，σ2Is the noise power.
For the above implementation, in detail, it is assumed that K independent far-field signals are incident on a Uniform Array (ULA) having M Array elements, and the incident direction of each far-field signal is θ1,...,θKThe true waveform of the kth far-field signal is sk(t), which may also be referred to as an ideal received signal. The actual received signal is at N sampling time t with same interval1,...,tNSampling is carried out, and the obtained matrix is a plurality of sampled snapshot signals X ═ X (t)1),...,x(tN)]. Wherein the content of the first and second substances,
Figure BDA0003016718860000052
v (t) white gaussian noise with zero mean; a (θ) represents a defect-free steering vector; and R isxx(θ)＝E[x(tN)xH(tN)]＝ASAH+RN；E[·]And (·)HRespectively representing a desired operator and a conjugate transform operator; a may be referred to as an array steering matrix and is defined as a ═ a (θ)1),a(θ2),...,a(θK)]S and RNThe covariance matrix and the noise matrix for an ideal received signal are defined as: s ═ E [ S (t) SH(t)]、RN＝E[v(t)vH(t)]. It should be noted that, since the noise follows a zero-mean gaussian distribution, the noise matrix can be expressed as: rN＝σ2I。
For the technical solution shown in fig. 1, in some possible implementations, the generating the received signal data in the beam space according to the covariance matrix of the array received signals and the beam pointing covariance matrix of each angle interval in the angle space includes:
dividing an angle space into a plurality of angle intervals, and determining a beam direction corresponding to each angle interval;
determining a covariance matrix corresponding to each angle interval according to the beam direction corresponding to each angle interval;
converting the covariance matrix corresponding to each angle interval into a column vector corresponding to each angle interval;
converting the covariance matrix of the array received signals into column vectors of the array received signals;
according to the n angleInterval-corresponding column vector r (gamma)n) And the column vector r (theta) of the array receiving signal is subjected to beam switching according to the following formula, and a beam space value of the array receiving signal in the nth angle interval is obtained:
gn＝rH(γn)r(θ)
wherein, γnA beam pointing value representing the nth angle interval,
and forming the signal data of the array receiving signals in the beam space according to the beam space values of all the angle intervals as follows:
g＝[g1,g2,...,gn]T
where T denotes the transpose operator.
For the above implementation, in detail, first, the angle space is divided into n parts, so as to obtain n angle intervals, and an angle at the center of each angle interval is taken as a beam direction corresponding to the angle interval, for example, the first angle space is [ β [ ]1,β2]Then the beam pointing direction of the first interval can be taken as gamma1＝(β1+β2) 2; if each angle interval is implemented in the same way, n beam pointing values (γ) representing the entire angle space will be obtained1,γ2,...,γn). Then, the covariance matrix of each beam pointing value can be calculated, and the covariance matrix R (γ) corresponding to each angle interval can be calculatedi). To obtain R (gamma)n) Then, it is converted into a column vector r (γ)i)＝rec{R(γi) And wherein rec {. denotes the convert matrix to column vector operator. Similarly, the covariance matrix R formed by the received signalsxxAfter (θ) is also converted to the column vector r (θ), the beam switching can be performed according to the following equation:
gi＝rH(γi)r(θ)
wherein, gnAnd (3) a beam space value corresponding to the ith angle interval, wherein i is 1.
Finally, the beam space values of all angle intervals are integrated together to form the beam spaceInter received signal data g ═ g1,g2,...,gn]TG is the received signal data of the beam space; this also means that the conversion of the received signal from the array element space to the beam space is complete.
For the technical solution shown in fig. 1, in some possible implementations, the method further includes completing a process of training a deep neural network DNN by using a training data set, and in some examples, the training process may include:
firstly, generating a training set and corresponding labels; in training, a dual-signal training set is typically employed to generalize to single-signal as well as three-signal scenarios. In some examples, some more specific angular interval Δ ═ { Δ ═ may be selected1,Δ2,...,ΔJAnd takes the input vector r (theta, delta)j) Representing the directions from theta and theta + deltajWherein theta is in the range of [ theta ]1,θI-Δj) Where J is 1., J denotes an angular interval of two signals, and I denotes the number of meshes of a spatial range divided by a unit angle. Based on the above, the training data set of the DNN can be represented as: gamma is ═ gamma1,...,ΓJ]Wherein
Figure BDA0003016718860000071
Representing the corresponding angle value of each grid; the corresponding tag is Ψ ═ Ψ1,...,ΨJ]Wherein
Figure BDA0003016718860000072
Figure BDA0003016718860000073
Indicating direction of origin
Figure BDA0003016718860000074
And
Figure BDA0003016718860000075
the two signals of (2) correspond to a tag vector.
Initializing a neural network which comprises a plurality of hidden layers and adopts a nonlinear activation function;
training the neural network by using a RMSProp optimizer through the training data set, and updating parameters of the neural network through back propagation in the training process so as to minimize reconstruction errors of a spatial spectrum; wherein the reconstruction error is expressed as:
Figure BDA0003016718860000076
y is the ideal output of the neural network,
Figure BDA0003016718860000077
representing an actual output of the neural network; the loss function used to characterize the minimized reconstruction error is the square of the spectral reconstruction error2Norm:
Figure BDA0003016718860000078
for the above example, in detail, in DNN, the embodiment of the present invention preferably adopts multiple hidden layers and adds a nonlinear activation function to enhance expressiveness, so as to achieve accurate direction of arrival estimation. In order to maintain the polarity of the input at each layer of the classifier, the nonlinear activation function used is a hyperbolic tangent function tanh (σ) [ tanh (σ) ]1),tanh(σ2),...,tanh(σ-1)]TWherein, in the step (A),
Figure BDA0003016718860000079
σ-1as the last element of the vector σ, σiRepresenting the ith element in the vector sigma. In the process of training the DNN, updating relevant parameters in the DNN through back propagation so as to minimize the reconstruction error of the spatial spectrum, wherein the variables are updated in an iterative manner as follows:
Figure BDA0003016718860000081
α may be an arbitrary weight matrix and offset vector, μ is a learning rate, αoldAnd alphanewRespectively, representing the values of the variables before and after the current update. In the embodiment of the present invention, it is,the reconstruction error can be expressed as:
Figure BDA0003016718860000082
where y is the ideal output data corresponding to the input, and
Figure BDA0003016718860000083
representing the actual output data. The loss function of the DNN may be set to l, the square of the spectral reconstruction error2Norm, i.e.
Figure BDA0003016718860000084
For the above example and detailed description, in the implementation, the training process of DNN is preferably done using a self-contained RMSProp Optimizer (Optimizer) in tensflow, as described above.
Based on the foregoing, in some examples, the inputting the received signal data in the beam space into a deep neural network DNN that has been trained using a training data set includes:
separating real and imaginary parts of signal data g of the array reception signals in a beam space;
and taking the real column vector obtained after separation as the input data of the DNN which finishes training by utilizing the training data set.
Based on the foregoing, in some examples, the obtaining DOA estimation value from the DNN output signal data by linear amplitude interpolation includes:
the DOA of each far-field signal is estimated by linear amplitude interpolation between two adjacent grids based on the signal data spectrum of the DNN output having non-zero positive values only on the grid adjacent to the actual signal direction.
Specifically, the linear amplitude difference value can extract a peak value from a frequency spectrum, record the index of a nonzero positive value, and count the number of nonzero positive value areas in an output frequency spectrum; secondly, angle estimation is carried out, the total energy of the current non-zero region (namely the sum of the output values of the current region) is calculated, the output spectrum values correspond to the angles of the angle space one by one, and the angle value obtained by the estimation of the current non-zero region is obtained through calculation (namely the product sum of the spectrum values in the non-zero region and the angle values of the angle space corresponding to the spectrum values is obtained, and the quotient is obtained through the total energy quotient of the current region); and finally, sequencing according to the total energy of each non-zero region, and selecting the first corresponding angle values with the maximum total energy corresponding to the number of signals as the estimated angle values.
For the above example, it is noted that the reconstructed spectrum has non-zero positive values only on the grid adjacent to the true signal direction, and the direction of each far-field signal can be estimated by linear amplitude interpolation between two adjacent grids.
For the explanation of the method for estimating the direction of arrival DOA in the beam space based on the deep neural network DNN provided by the embodiment of the present invention, the embodiment of the present invention verifies the effectiveness and feasibility of the method through specific experiments. In this experiment, a Uniform Linear Array (ULA) of 10 array elements was used to estimate the far-field signal direction incident from the spatial range [ -60 °, 60 °), i.e. M10, θ(0)＝-60°，θ(I)60 degrees. The adjacent array elements of the ULA are spaced by half the wavelength of the signal. The spatial spectrum is composed of 1 ° grids, so that a total of I is 120 grids, i.e.
Figure BDA0003016718860000091
The covariance vector r in the training dataset of DNN, and the vector in the test dataset were obtained from K400 snapshots. Based on the angle space, the space is divided into 20 parts, namely n is 20, the area range of each angle interval is 6 degrees, and the value of the central angle of each angle interval is taken as the beam direction, namely gamma1＝-57.5°,γ2＝-51.5°,...,γn＝56.5°。
For the training set of the DNN network, the [ -60 °, 60 °) space is also sampled at 1 ° intervals to obtain
Figure BDA0003016718860000092
And computing covariance vectors and associated labels. The signal-to-noise ratio of the snapshot is 10 db. Batch size 32, learning rate μ1Set up the iterations 0.001The number is 300, and each iteration of the data set is scrambled. The size of the input layer is 40 and the size of the hidden layer is chosen to be 4, respectively 80,160,320,240]The size of the output layer corresponds to the number of angle intervals, which is 120. The layers of the DNN network are Fully Connected Layers (FCL) as shown in fig. 2, and the sizes of the layers of the DNN network are shown in table 1.
TABLE 1
Layer name Size and breadth
Input layer
40
First hidden layer 80
Second layer of hidden layer 160
Third hidden layer 320
Fourth hidden layer 240
Output layer 120
Furthermore, all weights and biases of the DNN network are randomly initialized according to a uniform distribution between-0.1 and 0.1.
Based on the experimental conditions and the DNN network training set forth above, the direction of arrival DOA estimation method (may be abbreviated as Beam space) of the Beam space based on the DNN of the deep neural network according to the embodiment of the present invention is compared with the conventional DOA estimation method (may be abbreviated as Array element space) based on the Array element space, and the experimental results are shown in fig. 3 to fig. 7.
FIG. 3 compares the angle deviation of the Beam space method and the Array element space method for two signals in 31 and 41 directions under different SNR conditions. The abscissa represents the signal-to-noise ratio of the signal, varying from 0db to 10db, and the ordinate is the root mean square error of the DOA estimate. As can be seen from fig. 3, the estimation accuracy of both methods is significantly improved as the signal-to-noise ratio increases. In contrast, the DOA estimation method based on the DNN beam space according to the embodiment of the present invention is superior to the DOA estimation method based on the array element space. Therefore, the DOA estimation method based on the DNN beam space has strong adaptability in different signal-to-noise ratio environments.
FIG. 4 compares the angle estimation and estimation error of the Beam space method and the Array element space method at an angle interval of 9.4. The angular interval between the two signals does not appear in the training set and the direction of the second signal deviates from the pre-set training direction and the output spectral grid. Fig. 4 (a) and (b) show the estimated directions and the estimated errors of the two signals when the first signal direction is increased from-60 ° to 50 ° in steps of 1 ° in the virtual array beam space. In fig. 4 (c) and (d), the DOA estimation direction and estimation error based on the array element space in the same scene are plotted. The two methods are compared under the same training scene and testing scene, and the signal to noise ratio is 10 db. As can be seen from fig. 4, the estimation accuracy of both methods is high, the direction of arrival estimates match their true values well, and most of the estimation errors are less than 0.5 °. It can be seen that the performance of both methods in this scenario is almost equal.
Fig. 5 enlarges the angular separation of the two signals, 16.4 ° and 60 °, respectively. The angular direction of the first signal varies between-60 deg. and 43 deg. when the angular interval is 16.4, and the angular direction of the second signal varies between-60 deg. and 0 deg. when the interval is 60 deg.. Fig. 5 (a) and (c) show the estimated directions of the Beam space method at two angular intervals, respectively, and (b) and (d) show the estimated directions of the Array element space method at two angular intervals, respectively. It can be seen that both methods have almost the same distribution of estimated directions at the same angular interval, and their training and testing remain in a 10db environment. Therefore, both methods show good generalization ability to unknown scenes.
FIG. 6 shows the behavior of the Beam space method and the Array element space method when the test data contains different numbers of signals. Both approaches have been trained with array outputs in a dual signal scenario. The test scenario is still a 10db scenario, with the angular separation between adjacent signals in the three signal scenario being 20 °, the first signal varying between-60 ° and 40 °, and the first signal of the single signal scenario varying between-60 ° and 60 °. Fig. 6 (a) and (c) show the results of angle estimation of the Beam space method in a single-signal scene and a three-signal scene, and (b) and (d) show the results of estimation of the Array element space method. The experimental results show that in single-signal and double-signal scenes, the two methods have good generalization capability on unknown scenes; in a three-signal scene, when some incident signals are located at the edge of a filter sub-area, the corresponding direction-of-arrival estimation is deteriorated to some extent in accuracy and even disappears in the estimated spectrum. The DOA estimation method based on DNN beam space still shows satisfactory performance.
FIG. 7 compares the Beam space method and the Array element space method without error in training, but with error in four different cases taken into account in testing. Two signals with signal-to-noise ratios of 10db are assumed to impinge on the array from directions of 31.5 deg. and 41.5 deg., both directions being offset from the training and output spectral grids. Both methods obtain a very high accuracy of the direction of arrival estimate when there is no error. However, as array defects become more apparent, the direction of arrival estimation errors of both methods increase almost linearly. From the experimental results, it can be seen that the estimation deviation of the two methods is closer when the error intensity is smaller. With the gradual increase of the error intensity, the amplitude of the deviation rise estimated by the DOA estimation method based on the DNN beam space according to the embodiment of the present invention is significantly smaller than that of the DOA estimation method based on the array element space. The result is also foreseeable, and the DOA estimation method based on the DNN beam space in the embodiment of the invention carries out spatial spectrum estimation in the virtual array beam space, so that the DOA estimation method has strong adaptability to errors. Even if array errors are not considered in training, the method still has strong adaptability to the errors.
The following experiments and the description of the experimental results can be found: the DOA estimation method based on the DNN beam space has effectiveness and feasibility.
Based on the same inventive concept of the foregoing technical solution, referring to fig. 8, a DOA estimation apparatus 80 based on a DNN beam space according to an embodiment of the present invention is shown, where the apparatus 80 may include: an acquisition part 801, a generation part 802, a deep neural network DNN 803 and an interpolation part 804; wherein the content of the first and second substances,
the acquisition section 801 configured to acquire a covariance matrix of array reception signals;
the generating part 802 configured to generate the received signal data in the beam space according to the covariance matrix of the array received signals and the beam pointing covariance matrix of each angle interval in the angle space;
the DNN 803 configured to input and output reception signal data within the beam space;
the interpolation part 804 is configured to obtain DOA estimation values by linear amplitude interpolation from the signal data output by the DNN 803.
In the above scheme, the obtaining part 801 is configured to:
based on the incidence of K independent far-field signals to a uniform array with M array elements, the incidence direction of the kth far-field signal is set to be thetakThe k far field signal received by the uniform array is sk(t)；
Sampling the received far-field signal by N sampling moments with the same interval to obtain a snapshot signal X ═ X (t) of the array receiving signal1),...,x(tN)]；
Wherein the content of the first and second substances,
Figure BDA0003016718860000121
v (t) is zero-mean white gaussian noise, and a (θ) represents an ideal steering vector;
calculating a covariance matrix of the array received signals according to:
Rxx＝E[x(tN)xH(tN)]＝ASAH+RN
wherein, E [. C]And (·)HRespectively representing a desired operator and a conjugate transform operator; a is an array steering matrix, and A ═ a (θ)1),a(θ2),...,a(θk)](ii) a S and RNThe covariance matrix and the noise matrix, respectively, of the ideal received signal are defined as: s ═ E [ S (t) SH(t)]，RN＝E[v(t)vH(t)]R since the noise follows a zero mean Gaussian distributionN＝σ2I，σ2Is the noise power.
In the above scheme, the generating part 802 is configured to:
dividing an angle space into a plurality of angle intervals, and determining a beam direction corresponding to each angle interval;
determining a covariance matrix corresponding to each angle interval according to the beam direction corresponding to each angle interval;
converting the covariance matrix corresponding to each angle interval into a column vector corresponding to each angle interval;
converting the covariance matrix of the array received signals into column vectors of the array received signals;
according to the column vector r (gamma) corresponding to the n-th angle intervaln) And the column vector r (theta) of the array receiving signal is subjected to beam switching according to the following formula, and a beam space value of the array receiving signal in the nth angle interval is obtained:
gn＝rH(γn)r(θ)
wherein, γnA beam pointing value representing the nth angle interval,
and forming the signal data of the array receiving signals in the beam space according to the beam space values of all the angle intervals as follows:
g＝[g1,g2,...,gn]T
where T denotes the transpose operator.
In the above scheme, the generating part 802 is further configured to:
separating real and imaginary parts of signal data g of the array reception signals in a beam space;
and taking the real column vector obtained after separation as the input data of the DNN 803 which is trained by utilizing the training data set.
In the above solution, as shown in fig. 9, the apparatus 80 further includes a training section 805 configured to:
initializing a neural network which comprises a plurality of hidden layers and adopts a nonlinear activation function;
training the neural network by using a RMSProp optimizer through the training data set, and updating parameters of the neural network through back propagation in the training process so as to minimize reconstruction errors of a spatial spectrum; wherein the reconstruction error is expressed as:
Figure BDA0003016718860000131
y is the ideal output of the neural network,
Figure BDA0003016718860000132
representing an actual output of the neural network; the loss function used to characterize the minimized reconstruction error is the square of the spectral reconstruction error2Norm:
Figure BDA0003016718860000133
in the above scheme, the nonlinear activation function is a hyperbolic tangent function:
tanh(σ)＝[tanh(σ1),tanh(σ2),...,tanh(σ-1)]Twherein, in the step (A),
Figure BDA0003016718860000134
σ-1as the last element of the vector σ, σiRepresenting the ith element in the vector sigma.
In the above scheme, the interpolation part 804 is configured to:
based on the signal data spectrum output by the DNN 803 having non-zero positive values only on the grid adjacent to the actual signal direction, the DOA of each far-field signal is estimated by linear amplitude interpolation between two adjacent grids.
It is understood that in this embodiment, "part" may be part of a circuit, part of a processor, part of a program or software, etc., and may also be a unit, and may also be a module or a non-modular.
In addition, each component in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit. The integrated unit can be realized in a form of hardware or a form of a software functional module.
Based on the understanding that the technical solution of the present embodiment essentially or a part contributing to the prior art, or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium, and include several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (processor) to execute all or part of the steps of the method of the present embodiment. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
Therefore, the present embodiment provides a computer storage medium, which stores a DOA estimation program of DNN-based beam space, and when the DOA estimation program of DNN-based beam space is executed by at least one processor, the steps of the DOA estimation method of DNN-based beam space in the foregoing technical solution are implemented.
Referring to fig. 10, a specific hardware structure of a computing device 100 capable of implementing the DNN-based beam space DOA estimation apparatus 80 according to the embodiment of the present invention is shown, where the computing device 100 may be a wireless device, a mobile or cellular phone (including a so-called smart phone), a Personal Digital Assistant (PDA), a video game console (including a video display, a mobile video game apparatus, a mobile video conference unit), a laptop computer, a desktop computer, a television set-top box, a tablet computing apparatus, an e-book reader, a fixed or mobile media player, or the like. The computing device 100 includes: a receive array 1001, memory 1002, and processor 1003; the various components are coupled together by a bus system 1004. It is understood that the bus system 1004 is used to enable communications among the components. The bus system 1004 includes a power bus, a control bus, and a status signal bus in addition to a data bus. But for the sake of clarity the various busses are labeled in fig. 10 as the bus system 1004. Wherein the content of the first and second substances,
the receiving array 1001 is used for receiving an incident far-field signal;
the memory 1002 is used for storing a computer program capable of running on the processor 1003;
the processor 1003 is configured to, when running the computer program, execute the steps of the DOA estimation method based on the DNN beam space in the foregoing technical solution.
It is to be understood that the memory 1002 in embodiments of the present invention may be either volatile memory or nonvolatile memory, or may include both volatile and nonvolatile memory. The non-volatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable PROM (EEPROM), or a flash Memory. Volatile Memory can be Random Access Memory (RAM), which acts as external cache Memory. By way of illustration and not limitation, many forms of RAM are available, such as Static random access memory (Static RAM, SRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic random access memory (Synchronous DRAM, SDRAM), Double Data Rate Synchronous Dynamic random access memory (ddr Data Rate SDRAM, ddr SDRAM), Enhanced Synchronous SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory 1002 of the systems and methods described herein is intended to comprise, without being limited to, these and any other suitable types of memory.
And the processor 1003 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be implemented by integrated logic circuits of hardware or instructions in the form of software in the processor 1003. The Processor 1003 may be a general-purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, or discrete hardware components. The various methods, steps and logic blocks disclosed in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in the memory 1002, and the processor 1003 reads the information in the memory 1002 and performs the steps of the above method in combination with the hardware thereof.
It is to be understood that the embodiments described herein may be implemented in hardware, software, firmware, middleware, microcode, or any combination thereof. For a hardware implementation, the Processing units may be implemented within one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), general purpose processors, controllers, micro-controllers, microprocessors, other electronic units configured to perform the functions described herein, or a combination thereof.
For a software implementation, the techniques described herein may be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein. The software codes may be stored in a memory and executed by a processor. The memory may be implemented within the processor or external to the processor.
It should be understood that the above-mentioned exemplary technical solutions of the DOA estimation apparatus 80 and the computing device 100 based on the DNN beam space belong to the same concept as the technical solution of the foregoing DOA estimation method based on the DNN beam space, and therefore, the above-mentioned details that are not described in detail with respect to the technical solutions of the DOA estimation apparatus 80 and the computing device 100 based on the DNN beam space can be referred to the description of the technical solution of the foregoing DOA estimation method based on the DNN beam space. The embodiments of the present invention will not be described in detail herein.
It should be noted that: the technical schemes described in the embodiments of the present invention can be combined arbitrarily without conflict.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.

Claims (10)

1. A DOA (direction of arrival) estimation method of a beam space based on a Deep Neural Network (DNN), which is characterized by comprising the following steps:
acquiring a covariance matrix of array received signals;
generating received signal data in a beam space according to the covariance matrix of the array received signals and the beam pointing covariance matrix of each angle interval in the angle space;
inputting the received signal data in the beam space into a Deep Neural Network (DNN) which has completed training by utilizing a training data set;
and obtaining a DOA estimated value by linear amplitude interpolation of the signal data output by the DNN.
2. The method of claim 1, wherein obtaining a covariance matrix of array received signals comprises:
based on the incidence of K independent far-field signals to a uniform array with M array elements, the incidence direction of the kth far-field signal is set to be thetakThe k far field signal received by the uniform array is sk(t)；
Sampling the received far-field signal by N sampling moments with the same interval to obtain a snapshot signal X ═ X (t) of the array receiving signal1),...,x(tN)]；
Wherein the content of the first and second substances,
Figure FDA0003016718850000011
v (t) is zero-mean white gaussian noise, and a (θ) represents an ideal steering vector;
calculating a covariance matrix of the array received signals according to:
Rxx＝E[x(tN)xH(tN)]＝ASAH+RN
wherein, E [. C]And (·)HRespectively representing a desired operator and a conjugate transform operator; a is an array steering matrix, and A ═ a (θ)1),a(θ2),...,a(θk)](ii) a S and RNThe covariance matrix and the noise matrix, respectively, of the ideal received signal are defined as: s ═ E [ S (t) SH(t)]，RN＝E[v(t)vH(t)]R since the noise follows a zero mean Gaussian distributionN＝σ2I，σ2Is the noise power.
3. The method of claim 1, wherein generating the received signal data in the beam space based on the covariance matrix of the array received signals and the beam pointing covariance matrix for each angular interval in the angular space comprises:
dividing an angle space into a plurality of angle intervals, and determining a beam direction corresponding to each angle interval;
determining a covariance matrix corresponding to each angle interval according to the beam direction corresponding to each angle interval;
converting the covariance matrix corresponding to each angle interval into a column vector corresponding to each angle interval;
converting the covariance matrix of the array received signals into column vectors of the array received signals;
according to the column vector r (gamma) corresponding to the n-th angle intervaln) And the column vector r (theta) of the array receiving signal is subjected to beam switching according to the following formula, and a beam space value of the array receiving signal in the nth angle interval is obtained:
gn＝rH(γn)r(θ)
wherein, γnA beam pointing value representing the nth angle interval,
and forming the signal data of the array receiving signals in the beam space according to the beam space values of all the angle intervals as follows:
g＝[g1,g2,…,gn]T
where T denotes the transpose operator.
4. The method of claim 3, wherein the inputting the received signal data within the beam space into a Deep Neural Network (DNN) trained with a training data set comprises:
separating real and imaginary parts of signal data g of the array reception signals in a beam space;
and taking the real column vector obtained after separation as the input data of the DNN which finishes training by utilizing the training data set.
5. The method of claim 1, further comprising:
initializing a neural network which comprises a plurality of hidden layers and adopts a nonlinear activation function;
training the neural network by using a RMSProp optimizer through the training data set, and updating parameters of the neural network through back propagation in the training process so as to minimize reconstruction errors of a spatial spectrum; wherein the reconstruction error is expressed as:
Figure FDA0003016718850000021
y is the ideal output of the neural network,
Figure FDA0003016718850000022
representing an actual output of the neural network; the loss function used to characterize the minimized reconstruction error is the square of the spectral reconstruction error2Norm:
Figure FDA0003016718850000023
6. the method of claim 5, wherein the nonlinear activation function is a hyperbolic tangent function tanh (σ) ═ tanh (σ)1),tanh(σ2),...,tanh(σ-1)]TWherein, in the step (A),
Figure FDA0003016718850000031
σ-1as the last element of the vector σ, σiRepresenting the ith element in the vector sigma.
7. The method of claim 1 wherein obtaining DOA estimates from the DNN output signal data by linear amplitude interpolation comprises:
the DOA of each far-field signal is estimated by linear amplitude interpolation between two adjacent grids based on the signal data spectrum of the DNN output having non-zero positive values only on the grid adjacent to the actual signal direction.
8. An apparatus for DNN-based DOA estimation in beam space, the apparatus comprising: the method comprises the steps of obtaining a part, a generating part, a deep neural network DNN and an interpolation part; wherein the content of the first and second substances,
the acquisition section configured to acquire a covariance matrix of array reception signals;
the generation part is configured to generate the received signal data in the beam space according to the covariance matrix of the array received signals and the beam pointing covariance matrix of each angle interval in the angle space;
the DNN configured to input and output received signal data within the beam space;
the interpolation part is configured to obtain the DOA estimated value by linear amplitude interpolation of the signal data output by the DNN.
9. A computing device, comprising a receiving array, a memory, and a processor; wherein the content of the first and second substances,
the receiving array is used for receiving an incident far-field signal;
the memory for storing a computer program operable on the processor;
the processor, when running the computer program, is configured to perform the steps of the method for DOA estimation of a DNN-based beam space of any of claims 1 to 7.
10. A computer storage medium characterized in that the computer storage medium stores a DNN-based beam space DOA estimation program that when executed by at least one processor implements the DNN-based beam space DOA estimation method steps of any of claims 1 to 7.
CN202110390921.3A 2021-04-12 2021-04-12 DOA estimation method and device based on DNN (digital optical network) beam space and computer storage medium Active CN112946564B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110390921.3A CN112946564B (en) 2021-04-12 2021-04-12 DOA estimation method and device based on DNN (digital optical network) beam space and computer storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110390921.3A CN112946564B (en) 2021-04-12 2021-04-12 DOA estimation method and device based on DNN (digital optical network) beam space and computer storage medium

Publications (2)

Publication Number Publication Date
CN112946564A true CN112946564A (en) 2021-06-11
CN112946564B CN112946564B (en) 2024-02-02

Family

ID=76231780

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110390921.3A Active CN112946564B (en) 2021-04-12 2021-04-12 DOA estimation method and device based on DNN (digital optical network) beam space and computer storage medium

Country Status (1)

Country Link
CN (1) CN112946564B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115567091A (en) * 2022-09-21 2023-01-03 北京邮电大学 DOA estimation method and device based on TSS-LMS adaptive filtering
WO2023070994A1 (en) * 2021-10-27 2023-05-04 东南大学 Array super-resolution direction of arrival estimation method

Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2004088347A1 (en) * 2003-03-31 2004-10-14 Da Tang Mobile Communications Equipment Co., Ltd. Method for estimating the direction of arrival of fixation beam in space
CN1564494A (en) * 2004-04-01 2005-01-12 上海交通大学 Self-adaptive beam shaping method of beam space orthogonal FDM medulating system
US20050201204A1 (en) * 2004-03-11 2005-09-15 Stephane Dedieu High precision beamsteerer based on fixed beamforming approach beampatterns
CN106772225A (en) * 2017-01-20 2017-05-31 大连大学 Beam Domain DOA based on compressed sensing estimates
CN108089147A (en) * 2017-12-07 2018-05-29 西南电子技术研究所（中国电子科技集团公司第十研究所） Improved shortwave unit localization method
CN110967665A (en) * 2019-10-07 2020-04-07 西安电子科技大学 DOA estimation method of moving target echoes under multiple external radiation sources
CN111030952A (en) * 2019-12-25 2020-04-17 内蒙古大学 Beam space channel estimation method and system of millimeter wave system
CN111123202A (en) * 2020-01-06 2020-05-08 北京大学 Indoor early reflected sound positioning method and system
US20200202883A1 (en) * 2018-12-19 2020-06-25 Cirrus Logic International Semiconductor Ltd. Data aided method for robust direction of arrival (doa) estimation in the presence of spatially-coherent noise interferers
CN111366892A (en) * 2020-03-24 2020-07-03 西北工业大学 Massive MIMO DOA system based on neural network and implementation method
CN111458676A (en) * 2020-03-05 2020-07-28 北京邮电大学 Direction-of-arrival estimation method and device based on cascaded neural network
CN112565118A (en) * 2020-11-25 2021-03-26 哈尔滨工业大学 Channel estimation method for millimeter wave lens antenna array communication system
CN112600772A (en) * 2020-12-09 2021-04-02 齐鲁工业大学 OFDM channel estimation and signal detection method based on data-driven neural network

Patent Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2004088347A1 (en) * 2003-03-31 2004-10-14 Da Tang Mobile Communications Equipment Co., Ltd. Method for estimating the direction of arrival of fixation beam in space
US20050201204A1 (en) * 2004-03-11 2005-09-15 Stephane Dedieu High precision beamsteerer based on fixed beamforming approach beampatterns
CN1564494A (en) * 2004-04-01 2005-01-12 上海交通大学 Self-adaptive beam shaping method of beam space orthogonal FDM medulating system
CN106772225A (en) * 2017-01-20 2017-05-31 大连大学 Beam Domain DOA based on compressed sensing estimates
CN108089147A (en) * 2017-12-07 2018-05-29 西南电子技术研究所（中国电子科技集团公司第十研究所） Improved shortwave unit localization method
US20200202883A1 (en) * 2018-12-19 2020-06-25 Cirrus Logic International Semiconductor Ltd. Data aided method for robust direction of arrival (doa) estimation in the presence of spatially-coherent noise interferers
CN110967665A (en) * 2019-10-07 2020-04-07 西安电子科技大学 DOA estimation method of moving target echoes under multiple external radiation sources
CN111030952A (en) * 2019-12-25 2020-04-17 内蒙古大学 Beam space channel estimation method and system of millimeter wave system
CN111123202A (en) * 2020-01-06 2020-05-08 北京大学 Indoor early reflected sound positioning method and system
CN111458676A (en) * 2020-03-05 2020-07-28 北京邮电大学 Direction-of-arrival estimation method and device based on cascaded neural network
CN111366892A (en) * 2020-03-24 2020-07-03 西北工业大学 Massive MIMO DOA system based on neural network and implementation method
CN112565118A (en) * 2020-11-25 2021-03-26 哈尔滨工业大学 Channel estimation method for millimeter wave lens antenna array communication system
CN112600772A (en) * 2020-12-09 2021-04-02 齐鲁工业大学 OFDM channel estimation and signal detection method based on data-driven neural network

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
周书宇;黄宛宁;李崔春;: "一种基于RBF神经网络实现波束形成的方法", 计算机仿真, no. 10, pages 159 - 163 *
王荣秀;田雨波;张贞凯;: "基于局部保持投影和RBF神经网络的DOA估计", 科学技术与工程, no. 24, pages 7054 - 7058 *
谢坚;陈莹莹;: "基于TOFS的宽带信号波束空间DOA估计", 电子科技, no. 06, pages 60 - 63 *
邱佳锋;: "基于优化BM3D的毫米波大规模MIMO信道估计", 电信科学, no. 09 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023070994A1 (en) * 2021-10-27 2023-05-04 东南大学 Array super-resolution direction of arrival estimation method
CN115567091A (en) * 2022-09-21 2023-01-03 北京邮电大学 DOA estimation method and device based on TSS-LMS adaptive filtering

Also Published As

Publication number Publication date
CN112946564B (en) 2024-02-02

Similar Documents

Publication Publication Date Title
Vorobyov et al. Robust adaptive beamforming using worst-case performance optimization: A solution to the signal mismatch problem
Vu et al. Nonparametric missing sample spectral analysis and its applications to interrupted SAR
CN113176532B (en) DNN robustness DOA estimation method, device and storage medium based on array error and array element failure of beam space
CN108710103B (en) Strong and weak multi-target super-resolution direction finding and information source number estimation method based on sparse array
CN112946564B (en) DOA estimation method and device based on DNN (digital optical network) beam space and computer storage medium
CN111337873B (en) DOA estimation method based on sparse array
CN110888105B (en) DOA estimation method based on convolutional neural network and received signal strength
WO2012003018A1 (en) Method for low sidelobe operation of a phased array antenna having failed antenna elements
Sun et al. Real-valued DOA estimation with unknown number of sources via reweighted nuclear norm minimization
CN115356678B (en) DPNALM algorithm-based sparse array DOA estimation method
Zhang et al. An RCB-like steering vector estimation method based on interference matrix reduction
Chen et al. Robust DoA estimation using denoising autoencoder and deep neural networks
CN110954860B (en) DOA and polarization parameter estimation method
CN112766304A (en) Maneuvering array orientation estimation method based on sparse Bayesian learning
CN113687314B (en) Main lobe interference resistance method for radar
CN115372925A (en) Array robust adaptive beam forming method based on deep learning
CN117092585B (en) Single-bit quantized DoA estimation method, system and intelligent terminal
Yan et al. Computationally efficient direction of arrival estimation with unknown number of signals
Hosseini et al. DOA estimation using multiple measurement vector model with sparse solutions in linear array scenarios
Li et al. DOA estimation of underwater wideband signals based on improved OMP method
CN114563764A (en) Self-adaptive beam generation method and system
CN114355293A (en) Self-adaptive beam generation method and system
Kim et al. Fast Fourier-Domain Optimization Using Hybrid L 1−/L ${} _ {p} $-Norm for Autofocus in Airborne SAR Imaging
Rahmani et al. Two layers beamforming robust against direction‐of‐arrival mismatch
Khan et al. Beamspace matrix pencil method for direction of arrival estimation

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant