CN115494466A - Self-calibration method for distributed radar - Google Patents

Self-calibration method for distributed radar Download PDF

Info

Publication number
CN115494466A
CN115494466A CN202211158999.3A CN202211158999A CN115494466A CN 115494466 A CN115494466 A CN 115494466A CN 202211158999 A CN202211158999 A CN 202211158999A CN 115494466 A CN115494466 A CN 115494466A
Authority
CN
China
Prior art keywords
self
layer
calibration
data
radar
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.)
Pending
Application number
CN202211158999.3A
Other languages
Chinese (zh)
Inventor
陈鹏
李夏雨
赵熠明
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.)
Southeast University
CETC Yangzhou Baojun Electronic Co Ltd
Original Assignee
Southeast University
CETC Yangzhou Baojun Electronic Co Ltd
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 Southeast University, CETC Yangzhou Baojun Electronic Co Ltd filed Critical Southeast University
Priority to CN202211158999.3A priority Critical patent/CN115494466A/en
Publication of CN115494466A publication Critical patent/CN115494466A/en
Pending legal-status Critical Current

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
    • 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
    • G01S7/40—Means for monitoring or calibrating
    • G01S7/4052—Means for monitoring or calibrating by simulation of echoes
    • G01S7/4056—Means for monitoring or calibrating by simulation of echoes specially adapted to FMCW
    • 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
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]

Abstract

The invention discloses a self-calibration method for a distributed radar in the field of radar detection, which comprises the following steps: s1, establishing a self-calibration neural network model, which comprises a feature extraction layer and a full connection layer and is used for extracting corresponding feature data of a distance-angle two-dimensional graph and outputting a position estimation value according to the feature data; s2: constructing a distance-angle two-dimensional graph based on the distributed radar detection data, constructing a data set, and training a self-calibration network model by using the data set; s3: and inputting the distance-angle two-dimensional graph of the radar to be calibrated into the trained self-calibration network model, outputting calibration parameters, and realizing the self-adaptive estimation and calibration of the distributed radar. The self-calibration method can be used for quickly and adaptively estimating and calibrating the position of each radar, and the detection accuracy of the distributed radar is improved.

Description

Self-calibration method for distributed radar
Technical Field
The invention relates to the technical field of radars, in particular to a self-calibration method for a distributed radar.
Background
FMCW (Frequency Modulated Continuous Wave) radar has the advantages of simple structure, small size and easy use, and is widely applied to the fields of object detection, motion recognition, automatic driving and the like. Under the limit of policy and process, the transmission power of the FMCW radar is often relatively small, which results in a limited detection range of a single radar, and often has a poor detection effect on a target with a weak reflection capability of part of radar waves. The distributed radar system composed of multiple FMCW radars can realize the increase of detection area and the improvement of target detection performance. The spatial positions of each radar node constituting the distributed radar system are different from each other, and data of each radar node cannot be directly utilized. Therefore, the method has great significance in calibrating the data of the distributed radar network.
Patent No. 202111586447.8, which proposes a distributed radar network, achieves data synchronization for each radar node by connecting a control module to a communication bus in a plurality of radar transceivers, in conjunction with a clock signal. This approach, however, limits the layout and range of the distributed radar system while increasing hardware costs.
Patent No. 202110832937.5 proposes a distributed radar target positioning method, which realizes the solution of target position parameters by converting target position estimation into a nonlinear constraint optimization problem. However, this method needs to acquire the location parameters of different radar nodes in advance, but in some cases, the measurement difficulty of the location parameters is large.
Disclosure of Invention
The self-calibration method for the distributed radar solves the problem that calibration is difficult between distributed radar systems in the prior art, and achieves self-adaptive estimation and calibration of the position of each radar.
The embodiment of the application provides a self-calibration method for a distributed radar, which comprises the following steps:
s1, creating a self-calibration neural network model of the distributed radar, wherein the self-calibration neural network model comprises a feature extraction layer and a full connection layer, the feature extraction layer is used for extracting corresponding feature data of a distance-angle two-dimensional graph and inputting the feature data into the full connection layer, and the full connection layer is used for outputting a position estimation value according to the feature data;
s2: constructing the distance-angle two-dimensional graph based on distributed radar detection data, constructing a data set, and training the self-calibration network model by using the data set;
s3: and inputting the distance-angle two-dimensional graph of the radar to be calibrated into the trained self-calibration network model, and outputting calibration parameters by the self-calibration network model to realize the self-adaptive estimation and calibration of the distributed radar.
The beneficial effects of the above embodiment are as follows: the method comprises the steps of pre-establishing a self-calibration neural network model, inputting the distance-angle two-dimensional graph of the radar to be calibrated into the trained self-calibration network model, outputting calibration parameters quickly by the self-calibration network model, realizing self-adaptive calibration of the distributed radar, estimating and calibrating the position of each radar quickly and adaptively by the self-calibration method, and improving the detection precision of the distributed radar.
On the basis of the above embodiments, the present application can be further improved, specifically as follows:
in one embodiment of the present application, the feature extraction layer includes a convolution layer, a pooling layer, a batch normalization layer and an activation layer one, the full connection layer includes a flattening layer, a linear layer and an activation layer two, the convolution layer and the pooling layer are used for extracting the feature data, the batch normalization layer is used for normalizing the feature data, the activation layer one and the activation layer two are used for applying a nonlinear feature to the feature data, the flattening layer is used for converting the feature data into one-dimensional data, and the linear layer is used for performing linear transformation processing on the one-dimensional data. The distance-angle two-dimensional graph is subjected to feature extraction through a convolution layer and a pooling layer, is subjected to batch normalization through a batch normalization layer, is subjected to nonlinear processing of an activation layer I and then is input into a full-link layer, and finally is subjected to a flattening layer, a linear layer and an activation layer II to output an estimated value of the relative position of each radar.
In one embodiment of the present application, the convolutional layer is composed of C convolutional kernels, the size of each convolutional kernel is N × N, the step size of each convolutional kernel moving is S, the number of 0 padding used for the input data edge is P, and the feature size after convolutional layer processing is as follows
Figure BDA0003858600400000031
Wherein Input represents the size of Input data, output represents the size of Output data, and the number of channels of Output data is C. The convolutional layer is used for rapidly and accurately extracting the characteristic data corresponding to the distance-angle two-dimensional graph in each radar.
In one embodiment of the present application, the normalization processing formula of the batch normalization layer is as follows:
Figure BDA0003858600400000032
Figure BDA0003858600400000033
Figure BDA0003858600400000041
where X is all samples of the input, X i Represents one data in a sample, m represents the number of samples per each reading of the neural network, E (X) represents the mean of the samples, var (X) represents the variance of the samples, X represents the variance of the samples input Input data representing a batch normalization layer, X output Represents the output data of the batch normalization layer, and epsilon is a set value. Epsilon is a small constant, so that the abnormity that the denominator is 0 is avoided, and the batch normalization layer normalizes input data by calculating the mean value and the variance of the input data so as to solve the problem of unstable values in the self-calibration network and improve the stability and the convergence speed of the self-calibration network.
In one embodiment of the present application, the self-calibrating neural network model includes three of the feature extraction layers and one of the fully-connected layers. The output result of the third layer of feature extraction layer is used as the input of the full connection layer, so that the feature extraction precision is improved.
In one embodiment of the present application, in the step S2, the self-calibration network model is trained by using the data set, which includes:
dividing the data set into a training set and a verification set, and calculating a loss function of the self-calibration network model based on the training set, wherein the formula is as follows:
Figure BDA0003858600400000042
the Loss function is expressed by Loss, the output represents a calibration parameter output by the neural network model after prediction of a training set, the calibration parameter is the distance between radars, label represents the real value of the training set, n represents the number of samples input into the network at one time, sigma represents summation, and | represents the absolute value;
and optimizing self-calibration network parameters based on a gradient back propagation algorithm and the loss function, and obtaining the trained self-calibration network model after verification of the verification set.
In one embodiment of the present application, in the step S2, the data set is constructed as follows:
respectively determining the distance and the angle of a plurality of targets relative to a radar;
respectively calculating the distance and the angle of the plurality of targets relative to the other radar;
superposing the distances and angles of all the targets to obtain a distance-angle graph of each radar;
the distance-angle diagrams of different radars and the real distances between the radars corresponding to the distance-angle diagrams form a data set required by the training network together;
wherein the distance and angle of the target relative to different radars is calculated by the following formula:
Figure BDA0003858600400000051
Figure BDA0003858600400000052
wherein R and R 1 Representing the distance between the target and different radar, theta and theta 1 Representing the angle between the target and the different radars and L the distance between the radars.
Drawings
In order to more clearly illustrate the detailed description of the invention or the technical solutions in the prior art, the drawings used in the detailed description or the prior art description will be briefly described below. Throughout the drawings, like elements or portions are generally identified by like reference numerals. In the drawings, elements or portions are not necessarily drawn to scale.
FIG. 1 is a flow chart of a self-calibration method for a distributed radar provided by the present invention;
FIG. 2 is a schematic structural diagram of a self-calibrating neural network model provided by the present invention;
FIG. 3 is a first distance-angle plot used in the examples;
FIG. 4 is a distance-angle plot two used in the example;
fig. 5 is a schematic diagram of two different radars after the distance-angle maps are fused.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts belong to the protection scope of the present invention.
The invention provides a data fusion self-calibration method for a distributed radar. The radar equipment provided by the invention does not need a synchronization and positioning ranging function module.
Fig. 1 is an overall implementation flow of the method provided by the present invention.
The implementation flow of the invention comprises the following steps:
s1, a self-calibration neural network model of the distributed radar is created, wherein the self-calibration neural network model comprises a feature extraction layer and a full connection layer, the feature extraction layer is used for extracting corresponding feature data of a distance-angle two-dimensional graph and inputting the feature data into the full connection layer, and the full connection layer is used for outputting a position estimation value according to the feature data.
The self-calibration neural network model can be built through the neural network environments such as PyTorch and TensorFlow. The radar equipment is characterized by comprising functional modules such as an antenna module, a radio frequency module and a data processing module, and does not need the functions of signal synchronization, positioning and ranging and the like.
The self-calibration neural network model after parameter optimization can realize the prediction of the distance between the radars by inputting a distance-angle diagram of the radars. Assuming that the distance between the radar a and the radar B needs to be estimated currently, the parameters of the radar a and the radar B are set to be the same, the size of the distance-angle graph is X × Y, X represents the number of sampling points of the radar in the distance dimension, and Y represents the number of sampling points of the radar in the angle dimension, then the sizes of data input into the model by the radar a and the radar B should be N × 1 × X Y, respectively, where N is the size of data read by the model each time. Firstly, tensors formed by distance-angle graphs input by the radars A and B are spliced together along a second dimension, the size of spliced input data is N x 2 x y, and the spliced input data is used as original data for model subsequent processing.
The feature extraction layer of the self-calibration neural network model extracts features of raw data to be processed, and comprises a convolution layer, a pooling layer, a batch normalization layer and an activation layer I.
The convolution layer is composed of C convolution kernels, the size of each convolution kernel is N x N, the step length of each movement of the convolution kernel is S, the number of 0 padding used for the edge of input data is P, and the size of the characteristic graph after convolution layer processing is
Figure BDA0003858600400000071
In equation (1), input represents the size of Input data, output represents the size of Output data, and the number of channels of Output data is C.
The feature map processed by the convolutional layer is subjected to feature enhancement through the pooling layer again, so that the understanding of the model on the relative position information between the features is improved. The pooling layers are similar to the convolutional layers, each pooling layer is composed of pooling cores, each pooling core moves on the feature map, and the feature map is processed according to a certain rule. The feature map processed by the pooling layer has the same size calculation as the convolutional layer, and the only difference is that the number of channels of the input data and the output data of the pooling layer is the same.
The characteristic diagram output by the pooling layer is input into a batch normalization layer, and the batch normalization layer normalizes the input data by calculating the mean value and variance of the input data to solve the problem of unstable values in the network and improve the stability and convergence rate of the network。The calculation of the mean and variance is determined by the following formula
Figure BDA0003858600400000081
Figure BDA0003858600400000082
In the formulas (2) and (3), X is all samples input by the neural network at each time, and X i Representing one of all samples input by the neural network each time, m represents the number of samples read by the neural network each time, E (X) represents the mean value of the samples, var (X) represents the variance of the samples, and the output value of the characteristic diagram processed by the batch normalization layer is determined by the following formula
Figure BDA0003858600400000083
In the formula (4), X input Input data, X, representing a batch normalization layer output Represents the output data of the batch normalization layer, and epsilon is a set value. Epsilon is a small constant, e.g., 0.0001, and serves to avoid anomalies where the denominator is 0.
The activation layer applies a non-linear feature to the batch normalized feature map, in this example, using a modified linear unit (ReLU) as the activation function. The output of the activation layer is the result of the feature extraction layer. In this embodiment, the self-calibrating neural network model includes three feature convolution layers and one fully connected layer. And the output result of the third layer of feature extraction layer is used as the input of the full connection layer.
The full connection layer is composed of a flattening layer, a linear layer and an activation layer II. The flattening layer multi-dimensional input data is vectorized into one-dimensional data. The linear layer performs a linear transformation on the input data, and the process can be expressed by the following formula:
y＝xA T +b (5)
in the formula (5), y represents the output of the linear layer, x represents the input of the linear layer, and A represents a matrix composed of parameters of the linear layer, () T Representing transpose, b represents bias, which is a learnable parameter. In this example, activation layer two selects ReLU as the activation function. Fig. 2 is a network configuration diagram of the present example.
S2: and constructing a distance-angle two-dimensional graph based on the distributed radar detection data, constructing a data set, and training a self-calibration network model by using the data set.
The optimization of the network parameters is realized by a back propagation algorithm. The optimization goal of the network parameters is to reduce the difference between the predicted value and the true value of the self-calibrating network model. In this example, the distance between the predicted value and the true value is measured using the mean absolute value error as a loss function, which is calculated by the following formula:
Figure BDA0003858600400000101
in the formula (6), n represents the number of samples read by the neural network each time, output represents a calibration parameter output by the neural network model after prediction of the training set, the calibration parameter is the distance between radars, label represents a true value, loss represents a Loss function, and | | represents an absolute value. In this example, the network parameters are optimized using a random gradient descent method.
In the embodiment, a data set required by a self-calibration network model is generated by simulating echo signals of two radars with different positions when detecting a target, and the data set is divided into a training set and a verification set. In the generation process of the data set, firstly, the distance and the angle of a target relative to a certain radar are determined, then the distance and the angle of the target relative to another radar are calculated, the reflection sectional areas of the targets relative to different radars can be randomly generated, and the distance and the angle of the target relative to different radars are calculated by the following formulas:
Figure BDA0003858600400000102
Figure BDA0003858600400000103
in the formulae (7) and (8), R and R 1 Representing the distance between the target and different radar, theta and theta 1 Representing the angle between the target and the different radars and L the distance between the radars. A plurality of targets with different distances and angles relative to one radar are generated, and the distances and the angles of the targets relative to another radar at different radar intervals are calculated through formulas. And superposing all the targets to obtain a distance-angle diagram of each radar. The distance-angle maps of different radars and the real distances between the radars corresponding to the different radars together form a data set required by the training network.
And training the self-calibration network model through the training set, and verifying the self-calibration network model by combining the verification set after the loss function of the neural network is converged. After verification, the distances between different radars can be predicted. Fig. 3 and 4 are distance-angle diagrams used in this example. Fig. 5 is the result of fusing range-angle maps from two different radars.
S3: and inputting the distance-angle two-dimensional graph of the radar to be calibrated into the trained self-calibration network model, and outputting calibration parameters by the self-calibration network model to realize the self-adaptive estimation and calibration of the distributed radar.
The parameters represent the distance between the radars, and the conversion of the distance and the angle of the target relative to different radars can be realized by combining the output of the self-calibration network model and the formulas (7) and (8), so that the self-adaptive estimation and calibration of the distributed radars are realized.
Table 1 below shows the comparison of the performance of the prior papers with the process proposed in this patent. The method 1 is S.Li, J.Guo, R.xi, C.Duan, Z.ZHai, Y.He, peer track based Calibration for Multi-Radar Network, IEEE INFOCOM 2021, may 2021. The method observes the motion track of the tester in the overlapping observation area of two radars, and realizes the Calibration of Radar external parameters by simplifying the motion track into a straight line and calculating the difference of the slope and the position of the same track in the straight lines generated by different radars. The method 2 is S.Iwata, T.Koda and T.Sakamoto, multirad Data Fusion for multiplex Peer Measurement, IEEE Sensors Journal, vol.21, no.22, pp.25870-25879, november,2021. The method realizes the positioning of the same target by extracting the breath characteristics of a tester, the same target has different positions in different radars, and the affine transformation of the same target between different radars is calculated by combining with the Puruk analysis, thereby achieving the purpose of estimating the radar external reference. The method 3 is A.Shastri, M.Canil, J.Pegoraro, P.Casari and M.Rossi, mmSCALE, self-Calibration of mmWave Radar Networks from Human motion projects, 2022IEEE Radar conference, march,2022. The method considers the time synchronization between different Radar nodes, and realizes the automatic association of the track of the moving target by introducing the time Calibration error into the cost function. These methods all require on-line calibration by observing moving targets, which greatly limits the application scenarios.
TABLE 1 comparison of the Performance of the process proposed in this patent with other processes
Comparison method Method 1 Method 2 Method 3 The proposed method
Error (rice) 0.087 0.1 0.18 0.05
As is obvious from Table 1, the self-calibration accuracy of the distributed radar system adopting the method is obviously improved.
Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.

Claims (7)

1. A self-calibration method for a distributed radar, comprising the steps of:
s1, creating a self-calibration neural network model of the distributed radar, wherein the self-calibration neural network model comprises a feature extraction layer and a full connection layer, the feature extraction layer is used for extracting corresponding feature data of a distance-angle two-dimensional graph and inputting the feature data into the full connection layer, and the full connection layer is used for outputting a position estimation value according to the feature data;
s2: constructing the distance-angle two-dimensional graph based on distributed radar detection data, constructing a data set, and training the self-calibration network model by using the data set;
s3: and inputting the distance-angle two-dimensional graph of the radar to be calibrated into the trained self-calibration network model, and outputting calibration parameters by the self-calibration network model to realize the self-adaptive estimation and calibration of the distributed radar.
2. Self-calibration method according to claim 1, characterized in that: the feature extraction layer comprises a convolution layer, a pooling layer, a batch normalization layer and an activation layer I, the full connection layer comprises a flattening layer, a linear layer and an activation layer II, the convolution layer and the pooling layer are used for extracting feature data, the batch normalization layer is used for normalizing the feature data, the activation layer I and the activation layer II are used for applying nonlinear features to the feature data, the flattening layer is used for converting the feature data into one-dimensional data, and the linear layer is used for performing linear transformation processing on the one-dimensional data.
3. Self-calibration method according to claim 2, characterized in that: the convolutional layer is composed of C convolutional kernels, the size of each convolutional kernel is N x N, the step length of each movement of the convolutional kernels is S, the number of 0 padding used for the input data edge is P, and the size of the characteristic graph after being processed by the convolutional layer is N
Figure FDA0003858600390000011
Wherein Input represents the size of Input data, output represents the size of Output data, and the number of channels of Output data is C.
4. Self-calibration method according to claim 3, characterized in that: the normalization processing formula of the batch normalization layer is as follows:
Figure FDA0003858600390000021
Figure FDA0003858600390000022
Figure FDA0003858600390000023
where X is all samples of the input, X i Represents one data in a sample, m represents the number of samples read by the neural network per time, E (X) represents the sample mean, var (X) represents the sample variance, X represents the sample variance input Input data, X, representing a batch normalization layer output Represents the output data of the batch normalization layer, and epsilon is a set value.
5. Self-calibration method according to claim 4, characterized in that: the self-calibrating neural network model includes three of the feature extraction layers and one of the fully-connected layers.
6. Self-calibration method according to claim 1, characterized in that: in step S2, training the self-calibration network model using the data set specifically includes:
dividing the data set into a training set and a verification set, and calculating a loss function of the self-calibration network model based on the training set, wherein the formula is as follows:
Figure FDA0003858600390000024
the Loss function is expressed by Loss, the output expresses a calibration parameter output by the neural network model after prediction of a training set, label expresses a true value of the training set, n expresses the number of samples input into the network at one time, sigma expresses summation, and | expresses an absolute value;
and optimizing self-calibration network parameters based on a gradient back propagation algorithm and the loss function, and obtaining the trained self-calibration network model after verification of the verification set.
7. Self-calibration method according to claim 1, characterized in that: in step S2, the data set is constructed as follows:
respectively determining the distance and the angle of a plurality of targets relative to a radar;
respectively calculating the distance and the angle of the plurality of targets relative to the other radar;
superposing the distances and angles of all the targets to obtain a distance-angle graph of each radar;
the distance-angle diagrams of different radars and the real distances between the radars corresponding to the distance-angle diagrams form a data set required by a training network together;
wherein the distance and angle of the target relative to different radars is calculated by the following formula:
Figure FDA0003858600390000031
Figure FDA0003858600390000032
where R and R1 denote distances between the target and different radars, θ and θ 1 denote angles between the target and different radars, and L denotes a distance between the radars.
CN202211158999.3A 2022-09-22 2022-09-22 Self-calibration method for distributed radar Pending CN115494466A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211158999.3A CN115494466A (en) 2022-09-22 2022-09-22 Self-calibration method for distributed radar

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211158999.3A CN115494466A (en) 2022-09-22 2022-09-22 Self-calibration method for distributed radar

Publications (1)

Publication Number Publication Date
CN115494466A true CN115494466A (en) 2022-12-20

Family

ID=84470637

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211158999.3A Pending CN115494466A (en) 2022-09-22 2022-09-22 Self-calibration method for distributed radar

Country Status (1)

Country Link
CN (1) CN115494466A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116482618A (en) * 2023-06-21 2023-07-25 西安电子科技大学 Radar active interference identification method based on multi-loss characteristic self-calibration network

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116482618A (en) * 2023-06-21 2023-07-25 西安电子科技大学 Radar active interference identification method based on multi-loss characteristic self-calibration network
CN116482618B (en) * 2023-06-21 2023-09-19 西安电子科技大学 Radar active interference identification method based on multi-loss characteristic self-calibration network

Similar Documents

Publication Publication Date Title
CN110222748B (en) OFDM radar signal identification method based on 1D-CNN multi-domain feature fusion
CN110045323B (en) Matrix filling-based co-prime matrix robust adaptive beamforming algorithm
CN108696331B (en) Signal reconstruction method based on generation countermeasure network
CN109298383B (en) Mutual-prime array direction-of-arrival estimation method based on variational Bayes inference
CN110596687B (en) Riemann manifold-based single-base MIMO radar target detection method
CN110888105A (en) DOA estimation method based on convolutional neural network and received signal strength
CN115494466A (en) Self-calibration method for distributed radar
Efimov et al. Artificial neural network based angle-of-arrival estimator
CN111913155A (en) Two-dimensional DOA estimation method based on array radar
CN110413939B (en) Arrival angle estimation method based on atomic norm
Zhao et al. L1-norm constraint kernel adaptive filtering framework for precise and robust indoor localization under the internet of things
Zhang et al. Gaussian parameterized information aided distributed cooperative underwater positioning algorithm
Wisanmongkol et al. An ensemble approach to deep‐learning‐based wireless indoor localization
CN115508824A (en) Multi-target big data association fusion tracking method and system
Lin et al. Underwater source localization using time difference of arrival and frequency difference of arrival measurements based on an improved invasive weed optimization algorithm
CN113255171A (en) Direction finding error correction method and device based on transfer learning
Huang et al. Direction of arrival based on the multioutput least squares support vector regression model
CN113835107A (en) Signal processing method and device of array satellite navigation receiver and intelligent terminal
CN112162238B (en) Doppler frequency offset measuring method based on large-scale antenna array
Song et al. Algorithm based on the linear canonical transform for QFM signal parameters estimation
CN111901026B (en) Arrival angle estimation method in communication
Tamazirt et al. A novel advanced 3D‐IPS based on mmWaves and SOM‐MLP neural network
Zhang et al. Fine‐grained CSI fingerprinting for indoor localisation using convolutional neural network
Wang et al. Research on indoor positioning of power grid equipment based on deep learning
Lu et al. Efficient kernel based 3-D source localization via tensor completion

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