CN112180326B - Hierarchical distributed positioning and speed measuring method based on large-scale antenna array - Google Patents
Hierarchical distributed positioning and speed measuring method based on large-scale antenna array Download PDFInfo
- Publication number
- CN112180326B CN112180326B CN202010992818.1A CN202010992818A CN112180326B CN 112180326 B CN112180326 B CN 112180326B CN 202010992818 A CN202010992818 A CN 202010992818A CN 112180326 B CN112180326 B CN 112180326B
- Authority
- CN
- China
- Prior art keywords
- neural network
- scale antenna
- measurement
- positioning
- antenna array
- 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
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000005259 measurement Methods 0.000 claims abstract description 94
- 238000013528 artificial neural network Methods 0.000 claims abstract description 50
- 238000012549 training Methods 0.000 claims abstract description 42
- 238000003062 neural network model Methods 0.000 claims abstract description 41
- 238000003491 array Methods 0.000 claims abstract description 20
- 238000004364 calculation method Methods 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 6
- 230000002596 correlated effect Effects 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 41
- 230000000875 corresponding effect Effects 0.000 claims description 28
- 238000007781 pre-processing Methods 0.000 claims description 22
- 238000005070 sampling Methods 0.000 claims description 5
- 230000004931 aggregating effect Effects 0.000 claims 1
- 238000005516 engineering process Methods 0.000 description 4
- 238000000354 decomposition reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000000691 measurement method Methods 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000010363 phase shift Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
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
- G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
- G01S5/02—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using radio waves
-
- 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
- G01S11/00—Systems for determining distance or velocity not using reflection or reradiation
- G01S11/02—Systems for determining distance or velocity not using reflection or reradiation using radio waves
- G01S11/10—Systems for determining distance or velocity not using reflection or reradiation using radio waves using Doppler effect
-
- 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
- G01S3/02—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 using radio waves
- G01S3/14—Systems for determining direction or deviation from predetermined direction
- G01S3/143—Systems for determining direction or deviation from predetermined direction by vectorial combination of signals derived from differently oriented antennae
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- 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
Abstract
The invention discloses a layered distributed positioning and speed measuring method based on a large-scale antenna array, which comprises the following steps: obtaining a Doppler measurement depth neural network model and a speed measurement positioning depth neural network model through offline depth neural network training; real-time processing the multipath parallel received signals of the large-scale antenna array by using a Doppler measurement depth neural network model; converging Doppler measurement results measured by a plurality of distributed large-scale antenna arrays to a central station; the reporting result of each distributed large-scale antenna array is correlated at the central station; and sending the Doppler measurement results on the correlation into a training-completed speed measurement positioning depth neural network model to obtain target speed measurement and positioning results in real time. The invention can solve the problem of overhigh real-time calculation complexity in the system of a large-scale antenna array in the traditional positioning and speed measuring method.
Description
Technical Field
The invention relates to the technical field of information processing, in particular to a layered distributed positioning and speed measuring method based on a large-scale antenna array.
Background
Massive multiple-input multiple-output (Massive Multiple Input Multiple Output, massive MIMO) technology is one of the core technologies of the fifth generation (5th Generation,5G) and subsequent broadband mobile communications. Compared with the traditional MIMO, the technology adopts a complex three-dimensional space multiplexing technology by using a larger-scale three-dimensional antenna array, greatly improves the multiplexing capability of the air interface time and the frequency spectrum resource of the 5G system, and has higher communication capacity and better service capability.
Speed measurement and positioning are typical applications of wireless information systems, compared with conventional uniform linear arrays and uniform circular arrays, the large-scale array antenna improves the number of antenna units by an order of magnitude, the increase of the number of the antenna units enables a receiver to obtain higher signal space resolution, for example, resolution of azimuth and elevation three-dimensional space can be performed, and the large-scale antenna array is combined with spatial spectrums (such as MUSIC, ESPRIT and the like), so that the spatial resolution of better than 1 degree can be obtained in dense signal space, and on the basis, higher positioning precision can be obtained based on azimuth crossover.
Taking ESPRIT as an example: assuming that a radiation source located in the far field emits K narrowband signals of statistically independent and identical wavelength simultaneously, there are:
X 1 ＝A 1 S+N 1
X 2 ＝A 2 S+N 2 ＝A 1 ΦS+N 2
wherein: a is that 1 And A 2 Respectively representing array manifold for dividing the large-scale antenna array into two subarrays which are in offset relation; x is X 1 、X 2 、N 1 、N 2 Respectively representing incoming wave signal matrixes received by the two subarrays and Gaussian white noise matrixes corresponding to the two subarrays; s represents a signal vector; Φ represents the phase shift matrix of the incoming wave signal reaching the two subarrays,θ k the incidence angles (k=1, 2,3, …, K) of the respective incoming wave signals are shown.
For the entire array, the received incoming signal matrix X may be represented as follows:
the covariance matrix of the above formula can be obtained:
performing feature decomposition on the covariance matrix to obtain:
in U s Is a signal subspace formed by the feature vectors corresponding to the large feature values; u (U) N Is the noise subspace formed by the eigenvectors corresponding to the small eigenvalues.
There is a unique non-singular matrix T such that:
U s can be decomposed into:
thus, the first and second substrates are bonded together,wherein ψ=t -1 ΦT。
Solving linear equationsObtaining psi can obtain the arrival direction, and space positioning can be completed based on the intersection of azimuth.
In the positioning process, in order to obtain azimuth information, covariance is required to be calculated on a received signal, eigenvalue decomposition is performed, then a linear equation is solved, and in the process of solving the linear equation, matrix inversion is required. As the number of antenna elements of a large-scale antenna array increases by several orders of magnitude, the complexity of the matrix operation correspondingly increases by several orders of magnitude squared times. Therefore, the traditional positioning and speed measuring method has the problem of excessively high real-time calculation complexity under the system of a large-scale antenna array.
Disclosure of Invention
Aiming at the problems, the invention aims to provide a layered distributed positioning and speed measuring method based on a large-scale antenna array, so as to solve the problem that the real-time calculation complexity is too high in the traditional positioning and speed measuring method under the system of the large-scale antenna array.
A layering distributed positioning and speed measuring method based on a large-scale antenna array comprises the following steps:
obtaining a Doppler measurement depth neural network model and a speed measurement positioning depth neural network model through offline depth neural network training;
real-time processing the multipath parallel received signals of the large-scale antenna array by using a Doppler measurement depth neural network model;
converging Doppler measurement results measured by a plurality of distributed large-scale antenna arrays to a central station;
the reporting result of each distributed large-scale antenna array is correlated at the central station;
and sending the Doppler measurement results on the correlation into a training-completed speed measurement positioning depth neural network model to obtain target speed measurement and positioning results in real time.
According to the layered distributed positioning and speed measuring method based on the large-scale antenna array, the problems of speed measurement and positioning of targets are firstly resolved into the problems of Doppler measurement and speed measurement positioning, the problems of Doppler measurement and speed measurement positioning are respectively solved by using two types of neural networks (namely a Doppler measurement depth neural network model and a speed measurement positioning depth neural network model) which are connected in series, the problems are divided into two stages of offline training and online real-time calculation, the complex calculation process is placed in the offline training stage, the multipath parallel received signals of the large-scale antenna array are intelligently and real-time processed by using the offline training depth neural network model, the real-time calculation process is simpler and faster, and the problem of overhigh real-time calculation complexity under the system of the large-scale antenna array is effectively solved.
In addition, the layering distributed positioning and speed measuring method based on the large-scale antenna array can also have the following additional technical characteristics:
the step of obtaining the Doppler measurement depth neural network model through offline depth neural network training specifically comprises the following steps:
constructing a signal training set, wherein the signal training set comprises an input data matrix and an output result set, the input data matrix is an N multiplied by M matrix, N represents the number of radio frequency links of a large-scale antenna array, M represents the number of signal sampling points for one training, one output result set consists of 3 groups of data, the dimension of the 1 st group of data is 1 multiplied by 1, the Doppler frequency offset corresponding to the input data matrix is represented, the dimension of the 2 nd group of data is 1 multiplied by 3, the target relative speed corresponding to the input data matrix is represented, the dimension of the 3 rd group of data is 1 multiplied by 3, the initial position of a target corresponding to the input data matrix is represented, and one input data matrix corresponds to one output result set;
the signal training set groups, sequentially and orderly group two input nodes into a group, and input the group into a preprocessing unit for preprocessing;
preprocessing a signal training set, wherein each preprocessing unit corresponds to one of two paths of signals input to the preprocessing unit, delaying D sampling points, wherein D is a non-negative number, the other signal is directly connected, multiplying the directly connected signal and the delay signal, and solving the average value of multiplication results;
and training the deep neural network, namely after each input data matrix is processed by N/2 preprocessing units, inputting the N/2 preprocessing results obtained into the deep neural network to obtain the current output result of the deep neural network, and calculating the performance of the deep neural network by using the current output result of the deep neural network and the 1 st group of data in the output result set corresponding to the input data matrix, so that the training is continuously performed until the performance of the deep neural network reaches a preset threshold, and obtaining the Doppler measurement deep neural network model.
When the performance of the deep neural network is calculated, the normalized root mean square error is selected as the performance function.
The step of obtaining the speed measurement positioning deep neural network model through offline deep neural network training specifically comprises the following steps:
inputting training sets of K distributed large-scale antenna arrays into a trained Doppler measurement depth neural network model, wherein K is a positive integer, and obtaining K Doppler frequency offset values for each input data matrix;
the K Doppler frequency offset values are input into a speed measurement positioning deep neural network to obtain the current output result of the deep neural network, the 2 nd group data and the 3 rd group data in the output result set corresponding to the current output result and the input data matrix of the deep neural network are used for calculating the performance of the deep neural network, the performance function adopts normalized weighted root mean square error, and the calculation method is as follows:
wherein, (v) x,i ，v y,i ，v z,i ) And (p) x,i ，p y,i ，p z,i ) Respectively representing the 2 nd group data and the 3 rd group data in the output result set corresponding to the i-th input data matrix; (v' x,i ，v' y,i ，v' z,i ) And (p' x,i ，p' y,i ，p' z,i ) Respectively representing the current output obtained by inputting the ith input data matrix into the deep neural network; w (w) v And w p Weighting values, w, respectively representing the network performance corresponding to the 2 nd group data and the 3 rd group data in the output result set v And w p Positive and the sum of the two is 1; i represents the number of input data matrices;
and finally, training continuously until the performance of the deep neural network reaches a preset threshold, and obtaining the speed measurement and positioning deep neural network model.
The method specifically comprises the steps of processing multipath parallel received signals of a large-scale antenna array in real time by using a Doppler measurement depth neural network model, wherein the steps comprise:
and in the K distributed large-scale antenna arrays, K is a positive integer, the measured data are sent to a Doppler measurement depth neural network model after being grouped and preprocessed, the distributed measurement value of Doppler frequency offset is obtained in real time, and the measurement time and signal characteristic information are recorded.
The step of converging Doppler measurement results measured by the plurality of distributed large-scale antenna arrays to the central station specifically comprises the following steps:
and reporting the measured Doppler frequency offset, the corresponding recorded measurement time and signal characteristic information to a central site through a wired network or a wireless network.
In the step of associating the reporting results of each distributed large-scale antenna array by the central station, association rules are as follows: the time deviation of the measurement time is smaller than a preset threshold, and the normalized deviation of the signal characteristics is smaller than the preset threshold.
When the central station correlates the reporting results of the distributed large-scale antenna arrays, if the record of the signal characteristics is absent, the correlation is performed only according to the measurement time.
Drawings
The foregoing and/or additional aspects and advantages of embodiments of the invention will be apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:
fig. 1 is a flow chart of a distributed positioning and speed measuring method based on a large-scale antenna array according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a distributed positioning and velocity measurement method based on a large-scale antenna array according to an embodiment of the present invention;
fig. 3 is a schematic diagram of the operation performed on each preprocessing unit.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of 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, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1 and 2, a distributed positioning and velocity measurement method based on a large-scale antenna array according to an embodiment of the present invention includes steps S101 to S105.
S101, obtaining a Doppler measurement depth neural network model and a velocity measurement positioning depth neural network model through offline depth neural network training.
In this embodiment, the step S101 specifically includes the following steps:
the first step: when the large-scale antenna array is provided with N=32 radio frequency links and a test set is constructed, the aerial unmanned aerial vehicle and the large-scale antenna array are provided with 1000 different relative movement speeds and initial positions, each radio frequency link is continuously sampled by M=1024 points, 1000 input data matrixes with 32×1024 dimensions are recorded, corresponding Doppler frequency offset is calculated according to the set relative movement speeds and initial positions to form 1000 output result sets, 1000 input data matrixes with 32×1024 dimensions and 1000 output result sets form a training set, one output result set is composed of 3 groups of data, the dimension of the 1 st group of data is 1×1 and represents the Doppler frequency offset corresponding to the input data matrixes, the dimension of the 2 nd group of data is 1×3 and represents the target relative speed corresponding to the input data matrixes, the dimension of the 3 rd group of data is 1×3 and represents the target initial position corresponding to the input data matrixes, and one input data matrix corresponds to one output result set.
And a second step of: doppler measurement deep neural network training: firstly, each row of the input data matrix corresponds to a radio frequency link of the large-scale antenna array and is correspondingly input to an input node, namely, the 1 st row is input to the 1 st input node, the 2 nd row is input to the 2 nd input node, and so on;
then, training set grouping: two input nodes are orderly organized into a group in sequence and input into a preprocessing unit for preprocessing.
Next, training set preprocessing: as shown in fig. 3, each preprocessing unit delays one of two signals input to the preprocessing unit by d=0 sampling points, and the other signal is directly connected. Multiplying the through signal and the delay signal, and obtaining the average value of the multiplication result.
Finally, deep neural network training: after each input data matrix is processed by 16 preprocessing units, the obtained 16 preprocessing results are input into the deep neural network to obtain the current output result of the deep neural network, the current output result of the deep neural network and the 1 st group data in the output result set corresponding to the input data matrix are used for calculating the performance of the deep neural network, the normalized root mean square error is selected as the performance function, and the training is continuously carried out until the performance of the deep neural network reaches a preset threshold, and the Doppler measurement deep neural network model is obtained.
And a third step of: training a speed measurement and positioning neural network: at the central station, firstly, a training set of 3 distributed large-scale antenna arrays is input into a deep neural network model which is completed by the second training step, and 3 Doppler frequency offset values are obtained for each input data matrix.
Then, 3 Doppler frequency offset values are input into a speed measurement positioning deep neural network to obtain the current output result of the deep neural network, the 2 nd group data and the 3 rd group data in the output result set corresponding to the current output result and the input data matrix of the deep neural network are used for calculating the performance of the deep neural network, and the performance function adopts normalized weighted root mean square error, and the calculation method comprises the following steps:
wherein, (v) x,i ，v y,i ，v z,i ) And (p) x,i ，p y,i ，p z,i ) Respectively representing the 2 nd group data and the 3 rd group data in the output result set corresponding to the i-th input data matrix; (v' x,i ，v' y,i ，v' z,i ) And (p' x,i ，p' y,i ，p' z,i ) Respectively representing the current output obtained by inputting the ith input data matrix into the deep neural network; w (w) v And w p Weighting values, w, respectively representing the network performance corresponding to the 2 nd group data and the 3 rd group data in the output result set v And w p Positive and the sum of the two is 1; i=1000 represents the number of input data matrices.
And finally, training continuously until the performance of the deep neural network reaches a preset threshold, and obtaining the speed measurement and positioning deep neural network model.
S102, carrying out real-time processing on multipath parallel received signals of the large-scale antenna array by using a Doppler measurement depth neural network model.
In this embodiment, in 3 distributed large-scale antenna arrays, actual measurement data is sent to a trained deep neural network model after being grouped and preprocessed, so as to obtain a distributed measurement value of doppler frequency offset in real time, and measurement time and signal characteristic information are recorded.
S103, collecting Doppler measurement results measured by a plurality of distributed large-scale antenna arrays to a central station.
In this embodiment, the 3 distributed massive antenna arrays report the measured doppler frequency offset and the corresponding recorded measurement time and signal characteristic information (such as center frequency and modulation mode) to a central station through an ethernet or a 5G network, where the central station is one of the 3 distributed massive antenna arrays or other devices.
S104, the reporting results of the distributed large-scale antenna arrays are correlated at the central station.
In this embodiment, the central station correlates the doppler frequency offset measurement result according to the measurement time and the signal characteristics, and the correlation criteria are as follows:
the time deviation of the measured time is less than a preset threshold (e.g., 10 milliseconds) and the normalized deviation of the signal characteristic is less than a preset threshold (e.g., 2%).
In the above-mentioned association process, if the record of the signal characteristic is absent, the association is performed only according to the measurement time.
S105, the Doppler measurement results on the correlation are sent into a training-completed speed measurement positioning depth neural network model, and target speed measurement and positioning results are obtained in real time.
And sending the Doppler measurement results on the correlation into a training-completed speed measurement positioning depth neural network model, so that the target speed measurement and positioning results can be obtained in real time.
Practical tests show that the normalized mean square error of the positioning of the embodiment is better than 3R, wherein R represents the target distance.
In summary, according to the distributed positioning and speed measuring method based on the large-scale antenna array provided by the invention, the problems of target speed measurement and positioning are firstly decomposed into the problems of Doppler measurement and speed measurement positioning, the problems of Doppler measurement and speed measurement positioning are respectively solved by using two types of neural networks (namely a Doppler measurement depth neural network model and a speed measurement positioning depth neural network model) which are connected in series, the problems are divided into two stages of offline training and online real-time calculation, the complex calculation process is placed in the offline training stage, the multipath parallel received signals of the large-scale antenna array are intelligently and real-time processed by using the offline training depth neural network model, the real-time calculation process is simpler and quicker, and the problem of excessively high real-time calculation complexity under the system of the large-scale antenna array is effectively solved.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
While embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that: many changes, modifications, substitutions and variations may be made to the embodiments without departing from the spirit and principles of the invention, the scope of which is defined by the claims and their equivalents.
Claims (8)
1. The layered distributed positioning and speed measuring method based on the large-scale antenna array is characterized by comprising the following steps of:
obtaining a Doppler measurement depth neural network model and a speed measurement positioning depth neural network model through offline depth neural network training;
real-time processing is carried out on multiple paths of parallel received signals of the large-scale antenna array by using a Doppler measurement depth neural network model, so that Doppler measurement results are obtained;
converging Doppler measurement results measured by a plurality of distributed large-scale antenna arrays to a central station;
the reporting result of each distributed large-scale antenna array is correlated at the central station;
and sending the Doppler measurement results on the correlation into a training-completed speed measurement positioning depth neural network model to obtain target speed measurement and positioning results in real time.
2. The method for hierarchical distributed positioning and velocity measurement based on a large-scale antenna array according to claim 1, wherein the step of obtaining the doppler measurement depth neural network model through offline depth neural network training specifically comprises:
constructing a signal training set, wherein the signal training set comprises an input data matrix and an output result set, the input data matrix is an N multiplied by M matrix, N represents the number of radio frequency links of a large-scale antenna array, M represents the number of signal sampling points for one training, one output result set consists of 3 groups of data, the dimension of the 1 st group of data is 1 multiplied by 1, the Doppler frequency offset corresponding to the input data matrix is represented, the dimension of the 2 nd group of data is 1 multiplied by 3, the target relative speed corresponding to the input data matrix is represented, the dimension of the 3 rd group of data is 1 multiplied by 3, the initial position of a target corresponding to the input data matrix is represented, and one input data matrix corresponds to one output result set;
the signal training set groups, sequentially and orderly group two input nodes into a group, and input the group into a preprocessing unit for preprocessing;
preprocessing a signal training set, wherein each preprocessing unit corresponds to one of two paths of signals input to the preprocessing unit, delaying D sampling points, wherein D is a non-negative number, the other signal is directly connected, multiplying the directly connected signal and the delay signal, and solving the average value of multiplication results;
and training the deep neural network, namely after each input data matrix is processed by N/2 preprocessing units, inputting the N/2 preprocessing results obtained into the deep neural network to obtain the current output result of the deep neural network, and calculating the performance of the deep neural network by using the current output result of the deep neural network and the 1 st group of data in the output result set corresponding to the input data matrix, so that the training is continuously performed until the performance of the deep neural network reaches a preset threshold, and obtaining the Doppler measurement deep neural network model.
3. The method for hierarchical distributed positioning and velocity measurement based on a large-scale antenna array according to claim 2, wherein the performance function uses normalized root mean square error when calculating the performance of the deep neural network.
4. The method for hierarchical distributed positioning and velocity measurement based on a large-scale antenna array according to claim 2, wherein the step of obtaining the velocity measurement positioning deep neural network model through offline deep neural network training specifically comprises the following steps:
inputting training sets of K distributed large-scale antenna arrays into a trained Doppler measurement depth neural network model, wherein K is a positive integer, and obtaining K Doppler frequency offset values for each input data matrix;
the K Doppler frequency offset values are input into a speed measurement positioning deep neural network to obtain the current output result of the deep neural network, the 2 nd group data and the 3 rd group data in the output result set corresponding to the current output result and the input data matrix of the deep neural network are used for calculating the performance of the deep neural network, the performance function adopts normalized weighted root mean square error, and the calculation method is as follows:
wherein, (v) x,i ，v y,i ，v z,i ) And (p) x,i ，p y,i ，p z,i ) Respectively representing the 2 nd group data and the 3 rd group data in the output result set corresponding to the i-th input data matrix; (v' x,i ，v' y,i ，v' z,i ) And (p' x,i ，p' y,i ，p' z,i ) Respectively representing the current output obtained by inputting the ith input data matrix into the deep neural network; w (w) v And w p Weighting values, w, respectively representing the network performance corresponding to the 2 nd group data and the 3 rd group data in the output result set v And w p Positive and the sum of the two is 1; i represents the number of input data matrices;
and finally, training continuously until the performance of the deep neural network reaches a preset threshold, and obtaining the speed measurement and positioning deep neural network model.
5. The method for hierarchical distributed positioning and velocity measurement based on a large-scale antenna array according to claim 4, wherein the step of real-time processing the multipath parallel received signals of the large-scale antenna array by using a doppler measurement depth neural network model specifically comprises:
and in the K distributed large-scale antenna arrays, K is a positive integer, the measured data are sent to a Doppler measurement depth neural network model after being grouped and preprocessed, the distributed measurement value of Doppler frequency offset is obtained in real time, and the measurement time and signal characteristic information are recorded.
6. The method for hierarchical distributed positioning and velocity measurement based on a large-scale antenna array according to claim 5, wherein the step of aggregating the doppler measurement results measured by a plurality of distributed large-scale antenna arrays to a central station specifically comprises:
and reporting the measured Doppler frequency offset, the corresponding recorded measurement time and signal characteristic information to a central site through a wired network or a wireless network.
7. The method for hierarchical distributed positioning and speed measurement based on a large-scale antenna array according to claim 6, wherein in the step of associating the report results of each distributed large-scale antenna array by the central station, association rules are as follows: the time deviation of the measurement time is smaller than a preset threshold, and the normalized deviation of the signal characteristics is smaller than the preset threshold.
8. The method for hierarchical distributed positioning and speed measurement based on massive antenna arrays according to claim 7, wherein when the central station correlates the reported results of the distributed massive antenna arrays, the correlation is performed only according to the measurement time if the record of the signal characteristics is absent.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010992818.1A CN112180326B (en) | 2020-09-21 | 2020-09-21 | Hierarchical distributed positioning and speed measuring method based on large-scale antenna array |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010992818.1A CN112180326B (en) | 2020-09-21 | 2020-09-21 | Hierarchical distributed positioning and speed measuring method based on large-scale antenna array |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112180326A CN112180326A (en) | 2021-01-05 |
| CN112180326B true CN112180326B (en) | 2023-11-21 |
Family
ID=73956595
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010992818.1A Active CN112180326B (en) | 2020-09-21 | 2020-09-21 | Hierarchical distributed positioning and speed measuring method based on large-scale antenna array |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112180326B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015169709A1 (en) * | 2014-05-06 | 2015-11-12 | Sensurity Limited | Scanning bistatic perimeter intrusion detection system |
| WO2018095009A1 (en) * | 2016-11-22 | 2018-05-31 | 上海斐讯数据通信技术有限公司 | Multi-room positioning method based on wifi and server |
| CN108169708A (en) * | 2017-12-27 | 2018-06-15 | 中国人民解放军战略支援部队信息工程大学 | The direct localization method of modular neural network |
| CN109884581A (en) * | 2019-02-22 | 2019-06-14 | 南昌大学 | A kind of wide airspace passive direction finding method of Terahertz based on General Cell |
| CN110099017A (en) * | 2019-05-22 | 2019-08-06 | 东南大学 | The channel estimation methods of mixing quantization system based on deep neural network |
| WO2019161076A1 (en) * | 2018-02-19 | 2019-08-22 | Digital Global Systems, Inc. | Systems, methods, and devices for unmanned vehicle detection and threat management |
| CN110263855A (en) * | 2019-06-20 | 2019-09-20 | 深圳大学 | A method of it is projected using cobasis capsule and carries out image classification |
| CN110749859A (en) * | 2019-10-22 | 2020-02-04 | 清华大学 | Single base station array positioning method and device based on multiple carrier frequencies |
| CN111313943A (en) * | 2020-02-20 | 2020-06-19 | 东南大学 | Three-dimensional positioning method and device under deep learning assisted large-scale antenna array |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB0701869D0 (en) * | 2007-01-31 | 2007-03-14 | Cambridge Consultants | Adaptive radar |
| US20170208495A1 (en) * | 2013-03-15 | 2017-07-20 | DGS Global Systems, Inc. | Systems, methods, and devices for geolocation with deployable large scale arrays |
| US11609305B2 (en) * | 2019-12-27 | 2023-03-21 | Intel Corporation | Online radar phase calibration through static environment measurements |
-
2020
- 2020-09-21 CN CN202010992818.1A patent/CN112180326B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015169709A1 (en) * | 2014-05-06 | 2015-11-12 | Sensurity Limited | Scanning bistatic perimeter intrusion detection system |
| WO2018095009A1 (en) * | 2016-11-22 | 2018-05-31 | 上海斐讯数据通信技术有限公司 | Multi-room positioning method based on wifi and server |
| CN108169708A (en) * | 2017-12-27 | 2018-06-15 | 中国人民解放军战略支援部队信息工程大学 | The direct localization method of modular neural network |
| WO2019161076A1 (en) * | 2018-02-19 | 2019-08-22 | Digital Global Systems, Inc. | Systems, methods, and devices for unmanned vehicle detection and threat management |
| CN109884581A (en) * | 2019-02-22 | 2019-06-14 | 南昌大学 | A kind of wide airspace passive direction finding method of Terahertz based on General Cell |
| CN110099017A (en) * | 2019-05-22 | 2019-08-06 | 东南大学 | The channel estimation methods of mixing quantization system based on deep neural network |
| CN110263855A (en) * | 2019-06-20 | 2019-09-20 | 深圳大学 | A method of it is projected using cobasis capsule and carries out image classification |
| CN110749859A (en) * | 2019-10-22 | 2020-02-04 | 清华大学 | Single base station array positioning method and device based on multiple carrier frequencies |
| CN111313943A (en) * | 2020-02-20 | 2020-06-19 | 东南大学 | Three-dimensional positioning method and device under deep learning assisted large-scale antenna array |
Non-Patent Citations (7)
| Title |
|---|
| 5G高速移动系统中基于BP神经网络的多普勒频偏估计方法;王增浩;杨丽花;程露;张捷;梁彦;;电信科学(第04期);全文 * |
| Intelligent Beam Training for Millimeter-Wave Communications via Deep Reinforcement Learning;Jianjun Zhang;《2019 IEEE Global Communications Conference (GLOBECOM), Waikoloa, HI, USA》;全文 * |
| Three Passive TDOA-AOA Receivers-Based Flying-UAV Positioning in Extreme Environments;Zhenghai Wang;《IEEE SENSORS JOURNAL》;全文 * |
| 基于子孔径方位特性的穿墙雷达虚拟多视角成像方法;周辉林;《电讯技术》;全文 * |
| 探地雷达回波信号的特征提取和分类;周辉林;田茂;熊俊志;;电波科学学报(第04期);全文 * |
| 滑动大规模MIMO的信道特征分析和链路优化;杨锦;《中国优秀硕士学位论文全文数据库 信息科技辑》;全文 * |
| 穿墙雷达人体目标探测技术综述;金添;宋勇平;;电波科学学报(第04期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112180326A (en) | 2021-01-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110365612B (en) | Deep learning beam domain channel estimation method based on approximate message transfer algorithm | |
| Sun et al. | Fingerprint-based localization for massive MIMO-OFDM system with deep convolutional neural networks | |
| CN112073106B (en) | Millimeter wave beam prediction method and device, electronic device and readable storage medium | |
| CN102175989B (en) | Method for measuring incoherently distributed signal two-dimensional DOA (direction of arrival) | |
| CN110515038B (en) | Self-adaptive passive positioning device based on unmanned aerial vehicle-array and implementation method | |
| CN110113088B (en) | Intelligent estimation method for wave arrival angle of separated digital-analog hybrid antenna system | |
| CN110109051A (en) | The array with mutual coupling DOA estimation method of battle array is controlled based on frequency | |
| CN112995892B (en) | Large-scale MIMO fingerprint positioning method based on complex neural network | |
| CN112929962B (en) | Positioning method, positioning device, computer equipment and storage medium | |
| CN112699735B (en) | Near-field source positioning method based on deep learning | |
| CN112448742A (en) | Hybrid precoding method based on convolutional neural network under non-uniform quantization | |
| CN107104720A (en) | The relatively prime array adaptive beamforming method rebuild based on covariance matrix virtual Domain discretization | |
| CN111366892A (en) | Massive MIMO DOA system based on neural network and implementation method | |
| CN115102592B (en) | Multi-user MIMO beam forming method based on federal learning | |
| CN110907923A (en) | Bistatic EMVS-MIMO radar angle estimation algorithm and device based on parallel factor algorithm | |
| CN112180326B (en) | Hierarchical distributed positioning and speed measuring method based on large-scale antenna array | |
| CN112699992A (en) | Hybrid source positioning method based on convolutional neural network | |
| Kao et al. | AI-aided 3-D beamforming for millimeter wave communications | |
| CN116405348A (en) | Communication perception integrated system array antenna and 5G new air interface angle estimation method | |
| CN110208736A (en) | Non-circular signal uniform array direction of arrival angle estimation method based on fourth-order cumulant | |
| Sun et al. | Channel modeling by RBF neural networks for 5G mm-wave communication | |
| CN112162238B (en) | Doppler frequency offset measuring method based on large-scale antenna array | |
| WO2022166477A1 (en) | Positioning method and apparatus, base station, computer device, and storage medium | |
| CN111931121B (en) | Distributed positioning and speed measuring method based on large-scale antenna array | |
| CN115473553A (en) | Method and device for intelligently inhibiting beam squint and electronic equipment |
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 |