CN114217264B - Radio signal direction of arrival estimation and positioner based on degree of depth study - Google Patents
Radio signal direction of arrival estimation and positioner based on degree of depth study Download PDFInfo
- Publication number
- CN114217264B CN114217264B CN202111386282.XA CN202111386282A CN114217264B CN 114217264 B CN114217264 B CN 114217264B CN 202111386282 A CN202111386282 A CN 202111386282A CN 114217264 B CN114217264 B CN 114217264B
- Authority
- CN
- China
- Prior art keywords
- arrival
- deep learning
- signal
- positioning
- model
- 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
- 238000013135 deep learning Methods 0.000 claims abstract description 31
- 238000012545 processing Methods 0.000 claims abstract description 20
- 238000012549 training Methods 0.000 claims description 24
- 238000012795 verification Methods 0.000 claims description 15
- 239000011159 matrix material Substances 0.000 claims description 13
- 230000003044 adaptive effect Effects 0.000 claims description 9
- 238000000034 method Methods 0.000 claims description 9
- 238000002360 preparation method Methods 0.000 claims description 9
- 230000001629 suppression Effects 0.000 claims description 9
- 238000011176 pooling Methods 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Abstract
The application discloses a radio signal direction of arrival estimation and positioning device based on deep learning, which comprises a wireless signal receiving system, a data acquisition and signal processing system and a comprehensive positioning system; the wireless signal receiving system comprises four groups of omnidirectional uniform linear array antennas which are arranged in an E shape and are used for receiving electromagnetic waves emitted by a space wireless target information source; the data acquisition and signal processing system is used for receiving signals output by the array antenna, processing impulse noise contained in the signals, inputting the processed signals into the deep learning network model, and estimating the direction of arrival of a target information source; and the comprehensive positioning system realizes the accurate positioning of the radio target according to the direction-of-arrival estimation result obtained by the data acquisition and signal processing system and the geometric structure of the E-type array. The application improves the accuracy and the robustness of the estimation and the positioning of the direction of arrival of the radio target in the impulse noise environment.
Description
Technical Field
The invention relates to the technical field of signal processing, in particular to a radio signal direction-of-arrival estimation and positioning device based on deep learning.
Background
As an important technical means of passive radio target positioning, the problem of estimating the direction of arrival has received a lot of attention both in practical application and in theoretical research since the 60 th century. Most of the existing direction-of-arrival estimation methods are model driven, wherein the most representative is a super-resolution subspace-like direction-of-arrival estimation method, namely, an array output signal is used as input data of a direction-of-arrival estimation model to construct a covariance matrix, and singular value decomposition is carried out on the covariance matrix to further obtain a signal subspace and a noise subspace. And constructing a space spectrum through the noise subspace, and finally obtaining the direction of arrival estimation of the space target information source through spectrum peak search so as to realize positioning. Subspace-like methods are represented by the MUSIC method proposed by Schmidt and the ESPRIT method proposed by Roy. On this basis, an array element space ESPRIT algorithm, a beam space MUSIC algorithm, a root-MUSIC algorithm, a decorrelation MUSIC algorithm and the like are also sequentially proposed. However, these algorithms only aim at specific signal conditions, and do not establish a nonlinear mapping relationship between the array output signal and the spatial target source direction of arrival, which is low in accuracy and requires more data samples. And the above methods are all based on gaussian noise assumption and are not robust to impulse noise common in today's radio channels. I.e. impulse noise can seriously undermine the performance of the direction of arrival estimation and positioning of the above algorithm.
Disclosure of Invention
Aiming at the problems of estimating and positioning the direction of arrival of a wireless target signal source in a pulse noise environment, the invention provides a device for estimating and positioning the direction of arrival of a radio signal based on deep learning, which provides an adaptive pulse noise filter for suppressing pulse noise in an array output signal.
In order to achieve the above objective, the present application provides a radio signal direction of arrival estimation and positioning device based on deep learning, which comprises a wireless signal receiving system, a data acquisition and signal processing system and a comprehensive positioning system;
the wireless signal receiving system comprises four groups of omnidirectional uniform linear array antennas which are arranged in an E shape, wherein the included angles between the three groups of parallel array antennas and the fourth group of array antennas at the end part are between 60 and 90 degrees, and the three groups of parallel array antennas are used for receiving electromagnetic waves emitted by a spatial wireless target information source;
The data acquisition and signal processing system is used for receiving signals output by the array antenna, processing impulse noise contained in the signals, inputting the processed signals into the deep learning network model, and estimating the direction of arrival of a target information source;
And the comprehensive positioning system realizes the accurate positioning of the radio target according to the direction-of-arrival estimation result obtained by the data acquisition and signal processing system and the geometric structure of the E-type array.
Further, the data acquisition and signal processing system comprises a data preparation module, a model training module, a model verification module and a model application module.
Further, in the data preparation module, the specific implementation manner is as follows:
collecting electromagnetic wave signals emitted by a plurality of space known radio target information sources as samples by using an array antenna;
constructing a covariance matrix of the sample signal, and carrying out real number and serialization on the covariance matrix;
Labeling the serialized sample signals, and establishing a corresponding relation between the sample signals and the direction of arrival;
The marked sample signals are divided into two groups, one group is used as model training, and the other group is used as model verification.
Further, the step of carrying out real number and serialization on the covariance matrix specifically comprises the following steps:
Let the array output signal be x, covariance matrix be R x, and real-ize and serialize R x by equation (1):
Wherein, AndRespectively representing the real part and the imaginary part, (.) * represents the conjugate operation,
Further, in the model training module, the specific implementation manner is as follows:
performing impulse noise suppression on the acquired sample signal by using a front-end adaptive impulse noise filter;
Let the array output signal be x, define the adaptive impulse noise suppression factor as:
Wherein λ and ρ are adjustment coefficients, m x is the median value of x; weighting the signal x based on f aw (x) to construct an adaptive pulse noise filter, so as to realize the pulse noise suppression of the sample signal;
And taking the sample signal with the noise suppressed as a training sample signal, and performing repeated iterative training through an input layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, an unfolding layer, a full-connection layer and a softmax layer to obtain the deep learning network model with the direction of arrival estimated.
Further, the model verification module takes the model verification data generated by the data preparation module as the input of the deep learning network model, verifies the accuracy of the arrival direction estimation, and takes the deep learning network model generated by training as an application model if the accuracy meets the system requirement; if the system requirements are not met, continuing to iterate the training until the system requirements are met.
Further, the model application module is specifically implemented in the following manner:
the method comprises the steps of disposing a deep learning network model obtained after training and verification in an actual E-type array antenna scene;
After deployment, the data actually received by the array antenna is used as input to detect the direction of arrival of the space target source, and then the estimation of the direction of arrival of the space target source is completed.
Compared with the prior art, the technical scheme adopted by the application has the advantages that: the application provides the self-adaptive pulse noise filter to inhibit the pulse noise in the array output signal, so as to construct a multi-layer deep learning network model, and improve the accuracy and the robustness of the estimation and the positioning of the direction of arrival of the radio target in the pulse noise environment.
Drawings
FIG. 1 is a schematic diagram of a radio signal direction of arrival estimation and positioning device based on deep learning;
fig. 2 is a schematic diagram of a deep learning network model structure.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the particular embodiments described herein are illustrative only and are not intended to limit the application, i.e., the embodiments described are merely some, but not all, of the embodiments of the application.
Thus, the following detailed description of the embodiments of the application, as presented in the figures, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by a person skilled in the art without making any inventive effort, are intended to be within the scope of the present application.
Example 1
As shown in FIG. 1, the application provides a radio signal direction of arrival estimation and positioning device based on deep learning, which comprises a wireless signal receiving system, a data acquisition and signal processing system and a comprehensive positioning system. Wherein: the wireless signal receiving system comprises four groups of omnidirectional uniform linear array antennas which are arranged in an E shape, and the included angles between the three groups of parallel array antennas and the fourth group of array antennas are between 60 and 90 degrees and are used for receiving electromagnetic waves emitted by a space wireless target information source. The data acquisition and signal processing system is used for receiving signals output by the array antenna, processing impulse noise contained in the signals, inputting the processed signals into the deep learning network model, and estimating the direction of arrival of the target information source, so that the problem of poor accuracy of the traditional direction of arrival estimation method is solved. And the comprehensive positioning system realizes the accurate positioning of the radio target according to the direction-of-arrival estimation result obtained by the data acquisition and signal processing system and the geometric structure of the E-type array.
The data acquisition and signal processing system comprises a data preparation module, a model training module, a model verification module and a model application module, wherein:
1. The data preparation module is specifically implemented by the following steps:
1.1, using the array antenna shown in fig. 1 to collect electromagnetic wave signals emitted by a plurality of spatially known radio target sources as samples;
1.2 constructing a covariance matrix of the sample signal, and carrying out real number and serialization on the covariance matrix:
Let the array output signal be x, covariance matrix be R x, and real-ize and serialize R x by equation (1):
Wherein, AndRespectively representing the real part and the imaginary part, (.) * represents the conjugate operation,
1.3, Marking the serialized sample signals, and establishing a corresponding relation between the sample signals and the direction of arrival;
1.4 the marked sample signals are divided into two groups, one group is used for model training, and the other group is used for model verification.
2. In the model training module, the specific implementation mode is as follows:
the deep learning network model is shown in fig. 2, and firstly, pulse noise suppression is carried out on the acquired sample signals by utilizing a front-end self-adaptive pulse noise filter;
Let the array output signal be x, define the adaptive impulse noise suppression factor as:
Where λ and ρ are adjustment coefficients and m x is the median value of x. An adaptive pulse noise filter is constructed by weighting the signal x based on f aw (x), thereby realizing the pulse noise suppression of the sample signal.
And taking the sample signal with the noise suppressed as a training sample signal, and performing repeated iterative training through an input layer, a first convolution layer, a first pooling layer, a second convolution layer, a second minimization layer, an unfolding layer, a full-connection layer and a softmax layer to obtain a deep learning network model with the direction of arrival estimated.
3. The model verification module: taking the model verification data generated by the data preparation module as input of a deep learning network model, verifying the accuracy of the direction-of-arrival estimation, and taking the deep learning network model generated by training as an application model if the accuracy meets the system requirement; if the system requirements are not met, continuing to iterate the training until the system requirements are met.
4. The model application module is specifically realized by the following steps:
4.1, the deep learning network model obtained after training and verification is deployed in an actual E-type array antenna scene;
And 4.2, after deployment is completed, detecting the direction of arrival of the space target source by taking data actually received by the array antenna as input, and further completing estimation of the direction of arrival of the space target source.
And the comprehensive positioning system realizes the accurate positioning of the target according to the structural characteristics of the E-type array antenna and the direction-of-arrival estimation result of the space target source obtained by the deep learning network model estimation.
The foregoing descriptions of specific exemplary embodiments of the present invention are presented for purposes of illustration and description. It is not intended to limit the invention to the precise form disclosed, and obviously many modifications and variations are possible in light of the above teaching. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application to thereby enable one skilled in the art to make and utilize the invention in various exemplary embodiments and with various modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the claims and their equivalents.
Claims (5)
1. The radio signal direction of arrival estimation and positioning device based on deep learning is characterized by comprising a wireless signal receiving system, a data acquisition and signal processing system and a comprehensive positioning system;
The wireless signal receiving system comprises four groups of omnidirectional uniform linear array antennas which are arranged in an E shape, wherein the included angles between the three groups of parallel array antennas and the fourth group of array antennas at the end part are 60-90 degrees, and the three groups of parallel array antennas are used for receiving electromagnetic waves emitted by a spatial wireless target information source;
The data acquisition and signal processing system is used for receiving signals output by the array antenna, processing impulse noise contained in the signals, inputting the processed signals into the deep learning network model, and estimating the direction of arrival of a target information source;
The comprehensive positioning system realizes accurate positioning of a radio target according to the direction-of-arrival estimation result obtained by the data acquisition and signal processing system and the geometric structure of the E-type array;
the data acquisition and signal processing system comprises a data preparation module, a model training module, a model verification module and a model application module;
in the model training module, the specific implementation mode is as follows:
performing impulse noise suppression on the acquired sample signal by using a front-end adaptive impulse noise filter;
Let the array output signal be x, define the adaptive impulse noise suppression factor as:
Wherein λ and ρ are adjustment coefficients, m x is the median value of x; weighting the signal x based on f aw (x) to construct an adaptive pulse noise filter, so as to realize the pulse noise suppression of the sample signal;
And taking the sample signal with the noise suppressed as a training sample signal, and performing repeated iterative training through an input layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, an unfolding layer, a full-connection layer and a softmax layer to obtain the deep learning network model with the direction of arrival estimated.
2. The device for estimating and positioning a direction of arrival of a radio signal based on deep learning as claimed in claim 1, wherein the data preparation module is specifically implemented as follows:
collecting electromagnetic wave signals emitted by a plurality of space known radio target information sources as samples by using an array antenna;
constructing a covariance matrix of the sample signal, and carrying out real number and serialization on the covariance matrix;
Labeling the serialized sample signals, and establishing a corresponding relation between the sample signals and the direction of arrival;
The marked sample signals are divided into two groups, one group is used as model training, and the other group is used as model verification.
3. The device for estimating and positioning direction of arrival of radio signal based on deep learning as claimed in claim 2, wherein said real-izing and serializing covariance matrix is specifically:
let the array output signal be x, covariance matrix be R x, and real-ize and serialize R x by equation (1) (1):
Wherein, AndRespectively representing the real part and the imaginary part, (.) * represents the conjugate operation,
4. The device for estimating and positioning direction of arrival of radio signal based on deep learning according to claim 1, wherein the model verification module uses the model verification data generated by the data preparation module as input of the deep learning network model to verify the accuracy of the estimation of direction of arrival, and uses the deep learning network model generated by training as an application model if the accuracy meets the system requirement; if the system requirements are not met, continuing to iterate the training until the system requirements are met.
5. The device for estimating and positioning a direction of arrival of a radio signal based on deep learning as claimed in claim 1, wherein the model application module is specifically implemented by:
the method comprises the steps of disposing a deep learning network model obtained after training and verification in an actual E-type array antenna scene;
After deployment, the data actually received by the array antenna is used as input to detect the direction of arrival of the space target source, and then the estimation of the direction of arrival of the space target source is completed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111386282.XA CN114217264B (en) | 2021-11-22 | Radio signal direction of arrival estimation and positioner based on degree of depth study |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111386282.XA CN114217264B (en) | 2021-11-22 | Radio signal direction of arrival estimation and positioner based on degree of depth study |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114217264A CN114217264A (en) | 2022-03-22 |
| CN114217264B true CN114217264B (en) | 2024-06-28 |
Family
ID=
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110320490A (en) * | 2019-06-05 | 2019-10-11 | 大连理工大学 | A kind of radio wave arrival direction estimating method under the conditions of no direct signal |
| CN113109759A (en) * | 2021-04-10 | 2021-07-13 | 青岛科技大学 | Underwater sound array signal direction-of-arrival estimation method based on wavelet transformation and convolutional neural network |
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110320490A (en) * | 2019-06-05 | 2019-10-11 | 大连理工大学 | A kind of radio wave arrival direction estimating method under the conditions of no direct signal |
| CN113109759A (en) * | 2021-04-10 | 2021-07-13 | 青岛科技大学 | Underwater sound array signal direction-of-arrival estimation method based on wavelet transformation and convolutional neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104730491B (en) | A kind of virtual array DOA estimation method based on L-type battle array | |
| CN111123192B (en) | Two-dimensional DOA positioning method based on circular array and virtual extension | |
| CN109633522B (en) | Direction-of-arrival estimation method based on improved MUSIC algorithm | |
| CN106772224A (en) | A kind of L-type array estimating two-dimensional direction-of-arrival algorithm of use time frequency analysis | |
| CN109655799A (en) | The non-homogeneous thinned array direction-finding method of covariance matrix vectorization based on IAA | |
| CN104408278A (en) | A method for forming steady beam based on interfering noise covariance matrix estimation | |
| CN112130111B (en) | Single-snapshot two-dimensional DOA estimation method in large-scale uniform cross array | |
| CN108594164B (en) | Plane array DOA estimation method and equipment | |
| CN112051540B (en) | Quick high-precision direction finding method | |
| CN108398659B (en) | Direction-of-arrival estimation method combining matrix beam and root finding MUSIC | |
| CN112255629B (en) | Sequential ESPRIT two-dimensional incoherent distribution source parameter estimation method based on united UCA | |
| CN111965591A (en) | Direction-finding estimation method based on fourth-order cumulant vectorization DFT | |
| CN111610488A (en) | Method for estimating wave arrival angle of any array based on deep learning | |
| CN108761380A (en) | One kind is for carrying high-precision target Wave arrival direction estimating method | |
| CN108089146A (en) | A kind of high-resolution broadband Wave arrival direction estimating method to estimating angle error robust | |
| CN113466782B (en) | Mutual coupling correction DOA estimation method based on Deep Learning (DL) | |
| CN114217264B (en) | Radio signal direction of arrival estimation and positioner based on degree of depth study | |
| CN110208736A (en) | Non-circular signal uniform array direction of arrival angle estimation method based on fourth-order cumulant | |
| CN107241131B (en) | Beam forming method using signal non-circular characteristic | |
| CN114217264A (en) | Radio signal direction of arrival estimation and positioner based on degree of depth learning | |
| CN115310480A (en) | DOA estimation algorithm performance analysis method based on deep neural network | |
| CN112285642A (en) | Signal direction-of-arrival estimation method for non-overlapping optimized co-prime array | |
| CN104898094A (en) | Space-frequency two-dimensional beam forming method of compressed sampling array | |
| CN113625219B (en) | Vector conformal array DOA-polarization parameter joint estimation method based on sparse Bayesian learning | |
| CN116224218A (en) | CNN angle of arrival estimation method based on direct sampling data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |