CN114374927A - Multi-target direct positioning method based on block sparse Bayes - Google Patents

Multi-target direct positioning method based on block sparse Bayes Download PDF

Info

Publication number
CN114374927A
CN114374927A CN202111591347.4A CN202111591347A CN114374927A CN 114374927 A CN114374927 A CN 114374927A CN 202111591347 A CN202111591347 A CN 202111591347A CN 114374927 A CN114374927 A CN 114374927A
Authority
CN
China
Prior art keywords
signal
base station
target
grid
receiving base
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
CN202111591347.4A
Other languages
Chinese (zh)
Inventor
郭海召
窦修全
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.)
CETC 54 Research Institute
Original Assignee
CETC 54 Research Institute
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 CETC 54 Research Institute filed Critical CETC 54 Research Institute
Priority to CN202111591347.4A priority Critical patent/CN114374927A/en
Publication of CN114374927A publication Critical patent/CN114374927A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/023—Services making use of location information using mutual or relative location information between multiple location based services [LBS] targets or of distance thresholds
    • 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
    • G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
    • G06F17/141—Discrete Fourier transforms
    • G06F17/142—Fast Fourier transforms, e.g. using a Cooley-Tukey type algorithm
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/025—Services making use of location information using location based information parameters

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Data Mining & Analysis (AREA)
  • Signal Processing (AREA)
  • Computational Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Discrete Mathematics (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The invention relates to the field of signal processing, in particular to a multi-target direct positioning method based on combined TOA (time of arrival) and AOA (angle of arrival) of block sparse Bayes. The invention mainly comprises the following five parts: the method comprises the steps of receiving signal processing, dividing an interested area, constructing an over-complete dictionary, iteratively updating parameters and accumulating posterior means. The received signal processing is to prepare for iterative updating of parameters; the region of interest is divided so as to construct an over-complete dictionary on a specific grid; the overcomplete dictionary is constructed to provide a basis for parameter iterative updating in the next step; the parameter iteration updating is to recover the radiation source signal to obtain the posterior mean value of the signal; the posterior mean accumulation is to obtain the likelihood estimation value of each grid point, and the index of the corresponding dictionary is the estimation position of the radiation source. Compared with a DPD (digital Pre-distortion) method and a BOMP (Bomp processing) method, the method has higher resolution and better positioning performance, and is particularly suitable for the conditions of low signal-to-noise ratio and less snapshots.

Description

Multi-target direct positioning method based on block sparse Bayes
Technical Field
The invention relates to the field of signal processing, in particular to a multi-target direct positioning method based on joint TOA (time offset) and AOA (angle offset) of block sparse Bayes.
Background
In conventional object localization, two steps are typically used to locate the object. The method comprises the steps of firstly, estimating positioning parameters containing target position information from obtained received signals, and secondly, establishing and solving an equation by using the relation between the obtained positioning parameters and the target position. However, this approach easily proves to be suboptimal from an information theory point of view, since it ignores the correlation of information of the same target among different receiving base stations. The concept of direct positioning is to directly process the raw received signal data of each receiving base station without estimating the target positioning parameters, and construct an objective function only related to the target position to complete the estimation of the target position. Compared with the traditional two-step positioning method, the direct positioning method has the advantages of higher positioning precision, better performance under the condition of low SNR (signal noise ratio), no need of data association and the like.
Sparse Bayesian (sparse Bayesian learning) is a method applied in the field of compressed sensing for acquiring and reconstructing sparse or compressible signals. In target positioning, the position of a transmitted signal can be recovered by assuming that the signal obeys zero-mean Gaussian distribution and then utilizing a received signal and an overcomplete dictionary through Bayesian theory. Block Sparse Bayesian (Block Sparse Bayesian Learning) is a further Sparse Bayesian method that uses the time-domain correlation of a signal, and by dividing the time-domain correlated part of the signal into blocks, the correlation in the blocks is used when recovering the signal.
Disclosure of Invention
In order to avoid the problems in the prior art, the invention provides a new combined TOA (time offset) and AOA (angle offset) multi-target direct positioning method based on a block structure, and the BSBL parameter updating method is modified according to the characteristics of the signal attenuation factor and is applied to the positioning frame.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that:
a multi-target direct positioning method based on block sparse Bayes comprises the following steps:
s1, receiving signals emitted by a target area radiation source by a base station;
s2, carrying out fast Fourier transform on the received signal, and transforming the signal to a frequency domain;
s3, constructing an over-complete dictionary according to the divided target region; an array signal receiving model from each grid point divided by the overcomplete dictionary for the target area to a receiving base station;
s4, arranging the received signals transformed from the same array element into a column vector, dividing the column vector into a block, and constructing a compressed sensing model by using an over-complete dictionary;
s5, initializing sparse parameters of the transmitting signal blocks, correlation parameters of structures in the blocks and noise variance parameters, and calculating the mean value and variance of the posterior probability of the transmitting signals received by each receiving base station by using an over-complete dictionary and a compressed sensing model;
s6, updating the sparse parameters of the transmitting signal blocks, the correlation parameters of the structures in the blocks and the noise variance parameters by using a BSBL method;
s7, judging whether the mean value of the posterior probability of the emission signals before and after the parameter updating is smaller than a threshold value or reaches the maximum parameter updating iteration number, if so, entering the step S8, otherwise, returning to the step S5;
and S8, accumulating the mean values of the latest posterior probabilities obtained by different receiving base stations as likelihood estimated values, and obtaining the position of the radiation source target corresponding to the index in the over-complete dictionary.
Further, the fast fourier transform in step S2 is specifically:
Figure RE-GDA0003556353890000031
in the formula, ωkIs the frequency of the k-th discrete frequency bin,
Figure RE-GDA0003556353890000032
and
Figure RE-GDA0003556353890000033
respectively correspond to the received signal rl(t) nth target signal sn(t) and zero mean noise nl(t) data of the k-th discrete frequency point, al(pn) In response to the array from the nth target location to the/th receiving base station,
Figure RE-GDA0003556353890000034
βl,nfor complex scalar channel attenuation factors, τ, propagating from the nth target to the l received base station signall(pn) For the time delay from the nth target location to the/th receiving base station,
Figure RE-GDA0003556353890000035
is the phase delay, depending on the distance between the radiation source and the receiving array.
Further, the overcomplete dictionary in step S3 is specifically:
Figure RE-GDA0003556353890000036
Dl＝[al(p1),al(p2),…al(pq),…al(pQ)]
Figure RE-GDA0003556353890000037
in the formula philOvercomplete dictionary for the l-th receiving base station, DlAn array response matrix constructed for grid points demarcated in the target region of interest, al(pq) To be taken from the grid position pqArray response to the l-th receiving station, Q being the number of grid points, IKIs a unit matrix of K × K, K being the number of snapshots of the received signal, omegalFor the representation of the TOA information of the received signal in the frequency domain, ωkDigital angular frequency, sign for kth snapshot
Figure RE-GDA0003556353890000038
Representing the kronecker product, diag {. DEG } to construct a corresponding pairDiagonal matrix of corner elements, τl(pQ) To be slave grid position pQTime delay to the ith receiving base station.
Further, the expression of the compressed sensing model in step S4 is:
yl＝Φlxl+nl,l＝1,…,L,
Figure RE-GDA0003556353890000041
Figure RE-GDA0003556353890000042
Figure RE-GDA0003556353890000043
in the formula, ylRepresenting the column vectorisation, x, of the received signallRepresenting the column vectorisation of the transmitted signal, nlColumn vectorization, indicative of noise, vec (-) indicates reconnecting the matrix by column,
Figure RE-GDA0003556353890000044
shows the result of transforming the received signal of the kth snapshot of the ith base station to the frequency domain,
Figure RE-GDA0003556353890000045
representing the result of transforming the transmitted signal of the q-th grid to a representation in the frequency domain,
Figure RE-GDA0003556353890000046
representing the result of the transformation of the noise of the kth snapshot of the l receiving station onto the frequency domain, philDenotes an overcomplete dictionary of the L-th receiving base station, Q denotes the number of grid points, K denotes the number of snapshots of the received signal, and L denotes the number of receiving base stations.
Further, step S6 specifically includes the following steps:
s6-1, when the signal-to-noise ratio is higher than the set value, estimating the variance lambda of the noise after parameter updating*Comprises the following steps:
Figure RE-GDA0003556353890000047
wherein L represents the number of receiving base stations, M represents the number of array elements, K represents the number of snapshots of the received signal, ylRepresenting the column vectorisation of the received signal, philIs an overcomplete dictionary of the ith receiving base station, | · | | non-woven2Representing a vector two norm; tr (-) denotes the trace of the matrix, μl,xSum-sigmal,xRespectively mean value and variance of posterior probability of the transmitting signal;
when the signal-to-noise ratio is lower than the set value, the variance estimation lambda of the noise after parameter updating*Comprises the following steps:
Figure RE-GDA0003556353890000048
in the formula phil,qFor the array response corresponding to grid point q,
Figure RE-GDA0003556353890000051
the variance of the posterior probability of the transmitting signal corresponding to the grid point q;
s6-2, updating correlation structure parameter estimation B of transmitted signal*The formula is as follows:
Figure RE-GDA0003556353890000052
in the formula (I), the compound is shown in the specification,
Figure RE-GDA0003556353890000053
is the mean of the posterior probabilities of the transmitted signals corresponding to grid points Q, Q being the number of grid points, γqIs an unknown non-negative parameter for controlling the block sparsity of the q-th grid point transmitting signal;
s6-3, updating the q gridBlock sparsity estimation of point transmitted signals
Figure RE-GDA0003556353890000054
The formula is as follows:
Figure RE-GDA0003556353890000055
Figure RE-GDA0003556353890000056
in the formula, | represents the operation of taking the modulus value, and the upper label (·)oldParameter representing last parameter update, BqPositive definite matrix for representing correlation structure of acquired transmitting signal and setting certain threshold, after every update
Figure RE-GDA0003556353890000057
All the data are compared with the threshold, if the data are less than the threshold, the data are set to be 0, when the data are 0, the q grid corresponding to the data become 0, and the grid point has no transmitting target.
The invention has the beneficial effects that: the invention utilizes the characteristic that the channel attenuation only affects the amplitude of the signal and does not affect the sparsity and the correlation structure of the signal, so that the transmitting signal can be sparsely reconstructed and the target positioning can be completed under the condition that the compression dictionary contains unknown factor channel attenuation. In addition, the invention has higher positioning precision and resolution ratio under the conditions of lower signal-to-noise ratio and less sampling points.
Drawings
FIG. 1 is a schematic diagram of a simulation result of the method;
FIG. 2 is a graph of the variation of different positioning methods RMSE with signal to noise ratio;
fig. 3 shows the variation of different positioning methods RMSE with fast beat number.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
As shown in fig. 1, the direct positioning method based on block sparse bayes includes the following steps:
s1, constructing a signal propagation equation under the condition that N signals are emitted from a target area radiation source and are received by L receiving base stations with M receiving antennas;
the signals received by the receiving base station n are specifically:
Figure RE-GDA0003556353890000061
in the formula, rl(t) is the received signal; beta is al,nA complex scalar channel attenuation factor for signals propagating from the nth target to the lth receive base station; a isl(pn) Is an array response from the nth target location to the l receiving base station; snSignal form of nth target; tau isl(pn) Is the time delay from the nth target position to the ith receiving base station; n isl(t) is zero-mean white Gaussian noise.
Further, since the channel attenuation always occurs simultaneously with the signal and only affects the amplitude of the signal, the signal propagation equation in step S1 can be rewritten as:
Figure RE-GDA0003556353890000062
sl,n(t-τl(pn))＝βl,nsn(t-τl(pn))
s2, performing fast Fourier transform (fast Fourier transform) on the received signal to transform the signal to a frequency domain;
since the channel attenuation always occurs simultaneously with the signal, andonly the amplitude of the signal is affected, the channel attenuation and the signal may be combined and denoted as sl,n(t-τl(pn))＝βl,nsn(t-τl(pn) ); the received signal is subjected to fast Fourier transform to obtain:
Figure RE-GDA0003556353890000071
in the formula, ωkIs the frequency of the k-th discrete frequency bin,
Figure RE-GDA0003556353890000072
are respectively rl(t)， sn(t)，nl(t) data of the k-th discrete frequency bin.
Figure RE-GDA0003556353890000073
Is the phase delay, depending on the distance between the radiation source and the receiving array.
S3, constructing an over-complete dictionary according to the divided target region; an array signal receiving model from each grid point divided by the overcomplete dictionary for the target area to a receiving base station;
the overcomplete dictionary specifically comprises:
Figure RE-GDA0003556353890000074
Dl＝[al(p1),al(p2),…,al(pQ)]
Figure RE-GDA0003556353890000075
in the formula philAn overcomplete dictionary for the l-th receiving base station; dlAn array response matrix constructed for the grid points divided in the region of interest; a isl(pq) To be taken from the grid position pqArray response to the l-th receiving base station; q is one of grid pointsCounting; i isKAn identity matrix of size K × K; k is the number of snapshots of the received signal; omegalIs a representation of the TOA information of the received signal in the frequency domain; omegakA digital angular frequency for the kth snapshot; symbol
Figure RE-GDA0003556353890000076
Represents the kronecker product; diag {. is a diagonal matrix that constructs a corresponding diagonal.
S4, arranging the received signals transformed from the same array element into a column vector, dividing the column vector into a block, and constructing a compressed sensing model according to the block; the compressed sensing model specifically comprises:
yl＝Φlxl+nl,l＝1,…,L,
Figure RE-GDA0003556353890000081
Figure RE-GDA0003556353890000082
Figure RE-GDA0003556353890000083
in the formula, ylVectorizing a column of the received signal; x is the number oflVectorizing a column of the transmit signal; n islVectorizing columns for noise; vec (·) represents reconnecting the matrix by column;
Figure RE-GDA0003556353890000084
a result of converting the kth sample data of the ith base station receiving signal to a frequency domain;
Figure RE-GDA0003556353890000085
representing the result of transforming the transmission signal of the q-th grid to a frequency domain;
Figure RE-GDA0003556353890000086
the result of converting the kth sample data of the ith base station noise to the frequency domain representation is shown.
S5, calculating the mean and variance of the posterior probability of the transmitting signal received by each receiving base station;
the specific method for calculating the mean and the variance of the posterior probability of the transmitting signal obtained by each receiving base station comprises the following steps:
Figure RE-GDA0003556353890000087
Figure RE-GDA0003556353890000088
Σ0＝diag{γ1B,…,γQB}
wherein mul,xSum-sigmal,xRespectively mean value and variance of posterior probability of the transmitting signal; sigma0Is the variance of the transmitted signal; λ is the variance of gaussian white noise; i represents an identity matrix of corresponding size; gamma rayqIs an unknown non-negative parameter for controlling the block sparsity of the q-th grid point transmitting signal; b is a positive definite matrix used to obtain the correlation structure of the transmitted signal.
S6, updating the sparse parameters of the transmitting signal block, the correlation parameters of the structure in the block and the noise variance parameters by using a modified BSBL method; the method specifically comprises the following substeps:
s6-1, according to the formula:
Figure RE-GDA0003556353890000091
obtaining a variance estimate λ of the parameter updated noise*；||·||2Representing a vector two norm; tr (-) denotes the trace of the matrix;
when the signal-to-noise ratio is low, the formula can be changed to obtain better update performance:
Figure RE-GDA0003556353890000092
in the formula phil,qFor the array response corresponding to grid point q,
Figure RE-GDA0003556353890000093
the variance of the posterior probability of the transmitting signal corresponding to the grid point q;
s6-2, according to the formula:
Figure RE-GDA0003556353890000094
updating a correlation structure parameter estimate B of a transmitted signal*(ii) a Wherein
Figure RE-GDA0003556353890000095
Is the mean of the posterior probabilities of the transmitted signals corresponding to the grid point q.
S6-3, according to the formula:
Figure RE-GDA0003556353890000096
Figure RE-GDA0003556353890000097
updating block sparsity estimates for the qth mesh point transmitted signal
Figure RE-GDA0003556353890000098
| · | represents a modulo operation; superscript (·)oldRepresenting the parameter of the last parameter update. And set a certain threshold, after every update, gammaqAll the data are compared with the threshold, if the data are less than the threshold, the data are set to be 0, when the value of the data are 0, the q grid corresponding to the data will become 0, and the grid point can be considered to have no transmitting target.
S7, judging whether the mean value change of the posterior probability of the transmitting signal in the parameter updating of the two times is small enough or whether the mean value change reaches the maximum parameter updating iteration number, if so, entering the step 8, otherwise, returning to the step 5 to continuously iterate and update the parameters;
and S8, accumulating the latest posterior average values obtained by different receiving base stations as likelihood estimation values, wherein the position of the radiation source target corresponds to the index of the radiation source target in the dictionary.
In one embodiment of the present invention, the signal transmitted by the target is a Binary Phase Shift Keying (Binary Phase Shift Keying) signal. It is assumed that the transmitter and array have no relative motion during the observation interval and therefore no doppler shift.
The receiving of the signal received by the base station specifically includes:
Figure RE-GDA0003556353890000101
wherein r isl(t) is the received signal; beta is al,nA complex scalar channel attenuation factor for signals propagating from the nth target to the lth receive base station; a isl(pn) Is an array response from the nth target location to the l receiving base station; snSignal form of nth target; tau isl(pn) Is the time delay from the nth target position to the ith receiving base station; n isl(t) is zero mean noise.
Since the channel attenuation always occurs simultaneously with the signal and only affects the amplitude of the signal, the channel attenuation and the signal may be combined as denoted sl,n(t-τl(pn))＝βl,nsn(t-τl(pn) ); the received signal is subjected to fast Fourier transform to obtain:
Figure RE-GDA0003556353890000102
wherein ω iskIs the frequency of the k-th discrete frequency bin,
Figure RE-GDA0003556353890000103
are respectively rl(t)， sn(t)，nl(t) data of the k-th discrete frequency bin.
Figure RE-GDA0003556353890000104
Is the phase delay, depending on the distance between the radiation source and the receiving array. Further, in order to convert the received model of the signal into a common form of compressed sensing, the received signal is transformed as follows:
yl＝Φlxl+vl,l＝1,…,L
Figure RE-GDA0003556353890000111
Figure RE-GDA0003556353890000112
Figure RE-GDA0003556353890000113
Figure RE-GDA0003556353890000114
Dl＝[al(p1),al(p2),…,al(pQ)]
Figure RE-GDA0003556353890000115
where vec (-) represents the column-wise concatenation of the matrix into a column vector,
Figure RE-GDA0003556353890000116
represents the product of kronecker, IKRepresenting an identity matrix of size K × K, Q is the number of grids in the region of interest, and is also the number of atoms in the dictionary. PhilAn overcomplete dictionary for the l-th receiving base station; dlAn array response matrix constructed for the grid points divided in the region of interest; a isl(pq) To be slave grid pqArray response of location to the l-th receiving base station; omegalIs a representation of the TOA information of the received signal in the frequency domain; diag {. is a diagonal matrix that constructs a corresponding diagonal.
Suppose that all signal blocks xl,qAre all zero mean gaussian processes and obey the following distribution:
Figure RE-GDA0003556353890000117
Figure RE-GDA0003556353890000118
Σ0＝diag{γ1B,…,γQB}
wherein gamma isqIs an unknown non-negative parameter for controlling the block sparsity of the q-th grid point transmitting signal; b isqIs a positive definite matrix for obtaining the correlation structure of the transmitted signal, and to avoid overfitting, we denote all B with a unique Bq；Σ0Is xlThe variance of the prior probabilities.
The noise is assumed to also follow a zero-mean gaussian distribution:
Figure RE-GDA0003556353890000121
where λ is the variance of the noise and I is the identity matrix of the corresponding size, then xlThe posterior probability of (d) can be expressed as:
Figure RE-GDA0003556353890000122
Figure RE-GDA0003556353890000123
Figure RE-GDA0003556353890000124
wherein mul,xSum-sigmal,xRespectively, the mean and variance of the posterior probability of the transmitted signal.
The parameters can be updated to obtain corresponding estimates by using EM (Expectation-Maximization algorithm) algorithm:
Figure RE-GDA0003556353890000125
obtaining a variance estimate λ of the parameter updated noise*；||·||2Representing a vector two norm; tr (-) denotes the trace of the matrix;
when the signal-to-noise ratio is low, the formula can be changed to obtain better update performance:
Figure RE-GDA0003556353890000126
wherein phil,qAn array response for the corresponding grid point q;
Figure RE-GDA0003556353890000127
is the variance of the posterior probability of the transmitted signal corresponding to the grid point q.
Updating a correlation structure parameter estimate B of a transmitted signal*：
Figure RE-GDA0003556353890000128
Wherein
Figure RE-GDA0003556353890000129
Is the mean of the posterior probabilities of the transmitted signals corresponding to the grid point q.
Updating block sparsity estimates for the qth mesh point transmitted signal
Figure RE-GDA0003556353890000131
Figure RE-GDA0003556353890000132
Figure RE-GDA0003556353890000133
| · | represents a modulo operation; superscript (·)oldRepresenting the parameter of the last parameter update. And set a certain threshold, after every update, gammaqAll the data are compared with the threshold, if the data are less than the threshold, the data are set to be 0, when the value of the data are 0, the q grid corresponding to the data will become 0, and the grid point can be considered to have no transmitting target.
Finally calculating each grid point
Figure RE-GDA0003556353890000134
Wherein
Figure RE-GDA0003556353890000135
Is the posterior probability mean of the q-th grid point.
Figure RE-GDA0003556353890000136
As the likelihood estimation value of the grid point, the index of the dictionary corresponding to the maximum value is the estimated position of the radiation source.
In the specific implementation process, the processor is built as Intel (R) core (TM) i7-9700K CPU, the main frequency is 3.6GHz, and the memory is 32 GB; the software platform is as follows: a simulation experiment platform of a Windows 10 operating system and MATLAB 2016b, wherein 4 receiving base stations are respectively arranged at (-5, -5) km, (-5,5) km, (5, -5) km, and (5,5) km in the simulation experiment platform. Each receiving base station is provided with a uniform linear array with 10 array elements, and the distance between every two array elements is half of the wavelength of a target transmitting signal. The signal transmitted by each transmitter is set to bpsk (binary Phase shifting) signal with a bandwidth of 1MHz and a carrier frequency of 1 GHz.
As shown in FIG. 1, when two targets are located at (-30, -10) m and (20,10) m, respectively, the algorithm can accurately distinguish the two different targets. In comparing fig. 2 and fig. 3, two transmission targets are located at (0, -100) m and (0,100) m, respectively. Compared with DPD (direct Position determination) and BOMP (Block Orthogonal Matching pursuit) algorithms, the method can obtain better performance, under the simulation condition, the DPD algorithm cannot effectively distinguish two targets, and the performance obtained by the BOMP algorithm is inferior to that of the proposed method.
In summary, the present invention mainly includes the following five parts: the method comprises the steps of receiving signal processing, dividing an interested area, constructing an over-complete dictionary, iteratively updating parameters and accumulating posterior means. The received signal processing is to prepare for iterative updating of parameters; the region of interest is divided so as to construct an over-complete dictionary on a specific grid; the overcomplete dictionary is constructed to provide a basis for parameter iterative updating in the next step; the parameter iteration updating is to recover the radiation source signal to obtain the posterior mean value of the signal; the posterior mean accumulation is to obtain the likelihood estimation value of each grid point, and the index of the dictionary corresponding to the maximum value is the estimation position of the radiation source. Compared with a DPD (digital Pre-distortion) method and a BOMP (Bomp processing) method, the method has higher resolution and better positioning performance, and is particularly suitable for the conditions of low signal-to-noise ratio and less snapshots.

Claims (5)

1. A multi-target direct positioning method based on block sparse Bayes is characterized by comprising the following steps:
s1, receiving signals emitted by a target area radiation source by a base station;
s2, carrying out fast Fourier transform on the received signal, and transforming the signal to a frequency domain;
s3, constructing an over-complete dictionary according to the divided target region; an array signal receiving model from each grid point divided by the overcomplete dictionary for the target area to a receiving base station;
s4, arranging the received signals transformed from the same array element into a column vector, dividing the column vector into a block, and constructing a compressed sensing model by using an over-complete dictionary;
s5, initializing sparse parameters of the transmitting signal blocks, correlation parameters of structures in the blocks and noise variance parameters, and calculating the mean value and variance of the posterior probability of the transmitting signals received by each receiving base station by using an over-complete dictionary and a compressed sensing model;
s6, updating the sparse parameters of the transmitting signal blocks, the correlation parameters of the structures in the blocks and the noise variance parameters by using a BSBL method;
s7, judging whether the mean value of the posterior probability of the emission signals before and after the parameter updating is smaller than a threshold value or reaches the maximum parameter updating iteration number, if so, entering the step S8, otherwise, returning to the step S5;
and S8, accumulating the mean values of the latest posterior probabilities obtained by different receiving base stations as likelihood estimated values, and obtaining the position of the radiation source target corresponding to the index in the over-complete dictionary.
2. The multi-target direct positioning method based on block sparse Bayes as claimed in claim 1, wherein the fast Fourier transform in step S2 specifically comprises:
Figure FDA0003429209920000011
in the formula, ωkIs the frequency of the k-th discrete frequency bin,
Figure FDA0003429209920000021
and
Figure FDA0003429209920000022
respectively correspond to the received signal rl(t) nth target signal sn(t) and zero mean noise nl(t) data of the k-th discrete frequency point, al(pn) In response to the array from the nth target location to the/th receiving base station,
Figure FDA0003429209920000023
βl,nfor propagation from the nth objectComplex scalar channel attenuation factor, tau, to the ith received base station signall(pn) For the time delay from the nth target location to the/th receiving base station,
Figure FDA0003429209920000024
is the phase delay, depending on the distance between the radiation source and the receiving array.
3. The multi-target direct positioning method based on block sparse Bayes as claimed in claim 1, wherein the overcomplete dictionary in step S3 is specifically:
Figure FDA0003429209920000025
Dl＝[al(p1),al(p2),…al(pq),…al(pQ)]
Figure FDA0003429209920000026
in the formula philOvercomplete dictionary for the l-th receiving base station, DlAn array response matrix constructed for grid points demarcated in the target region of interest, al(pq) To be taken from the grid position pqArray response to the l-th receiving station, Q being the number of grid points, IKIs a unit matrix of K × K, K being the number of snapshots of the received signal, omegalFor the representation of the TOA information of the received signal in the frequency domain, ωkDigital angular frequency, sign for kth snapshot
Figure FDA0003429209920000027
Representing the kronecker product, diag {. is a diagonal matrix, τ, that constructs a corresponding diagonal elementl(pQ) To be slave grid position pQTime delay to the ith receiving base station.
4. The block sparse Bayesian-based multi-target direct localization method as claimed in claim 1, wherein the expression of the compressed sensing model in step S4 is as follows:
yl＝Φlxl+nl,l＝1,…,L,
Figure FDA0003429209920000028
Figure FDA0003429209920000031
Figure FDA0003429209920000032
in the formula, ylRepresenting the column vectorisation, x, of the received signallRepresenting the column vectorisation of the transmitted signal, nlColumn vectorization, indicative of noise, vec (-) indicates reconnecting the matrix by column,
Figure FDA0003429209920000033
shows the result of transforming the received signal of the kth snapshot of the ith base station to the frequency domain,
Figure FDA0003429209920000034
representing the result of transforming the transmitted signal of the q-th grid to a representation in the frequency domain,
Figure FDA0003429209920000035
representing the result of the transformation of the noise of the kth snapshot of the l receiving station onto the frequency domain, philDenotes an overcomplete dictionary of the L-th receiving base station, Q denotes the number of grid points, K denotes the number of snapshots of the received signal, and L denotes the number of receiving base stations.
5. The multi-target direct positioning method based on block sparse Bayes as claimed in claim 1, wherein step S6 specifically comprises the following steps:
s6-1, when the signal-to-noise ratio is higher than the set value, estimating the variance lambda of the noise after parameter updating*Comprises the following steps:
Figure FDA0003429209920000036
wherein L represents the number of receiving base stations, M represents the number of array elements, K represents the number of snapshots of the received signal, ylRepresenting the column vectorisation of the received signal, philIs an overcomplete dictionary of the ith receiving base station, | · | | non-woven2Representing a vector two norm; tr (-) denotes the trace of the matrix, μl,xSum-sigmal,xRespectively mean value and variance of posterior probability of the transmitting signal;
when the signal-to-noise ratio is lower than the set value, the variance estimation lambda of the noise after parameter updating*Comprises the following steps:
Figure FDA0003429209920000037
in the formula phil,qFor the array response corresponding to grid point q,
Figure FDA0003429209920000038
the variance of the posterior probability of the transmitting signal corresponding to the grid point q;
s6-2, updating correlation structure parameter estimation B of transmitted signal*The formula is as follows:
Figure FDA0003429209920000041
in the formula (I), the compound is shown in the specification,
Figure FDA0003429209920000042
to correspond toMean value of posterior probability of transmitted signal of grid point Q, where Q is the number of grid points, γqIs an unknown non-negative parameter for controlling the block sparsity of the q-th grid point transmitting signal;
s6-3, updating block sparsity estimation of q grid point emission signal
Figure FDA0003429209920000043
The formula is as follows:
Figure FDA0003429209920000044
Figure FDA0003429209920000045
in the formula, | represents the operation of taking the modulus value, and the upper label (·)oldParameter representing last parameter update, BqPositive definite matrix for representing correlation structure of acquired transmitting signal and setting certain threshold, after every update
Figure FDA0003429209920000046
All the data are compared with the threshold, if the data are less than the threshold, the data are set to be 0, when the data are 0, the q grid corresponding to the data become 0, and the grid point has no transmitting target.
CN202111591347.4A 2021-12-23 2021-12-23 Multi-target direct positioning method based on block sparse Bayes Pending CN114374927A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111591347.4A CN114374927A (en) 2021-12-23 2021-12-23 Multi-target direct positioning method based on block sparse Bayes

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111591347.4A CN114374927A (en) 2021-12-23 2021-12-23 Multi-target direct positioning method based on block sparse Bayes

Publications (1)

Publication Number Publication Date
CN114374927A true CN114374927A (en) 2022-04-19

Family

ID=81142101

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111591347.4A Pending CN114374927A (en) 2021-12-23 2021-12-23 Multi-target direct positioning method based on block sparse Bayes

Country Status (1)

Country Link
CN (1) CN114374927A (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107703477A (en) * 2017-09-11 2018-02-16 电子科技大学 The steady broadband array signal Wave arrival direction estimating method of standard based on block management loading
US20200057163A1 (en) * 2018-08-16 2020-02-20 Movano Inc. Bayesian Geolocation and Parameter Estimation by Retaining Channel and State Information
CN111007457A (en) * 2018-10-08 2020-04-14 哈尔滨工业大学 Radiation source direct positioning method based on block sparse Bayesian model
CN111007458A (en) * 2018-10-08 2020-04-14 哈尔滨工业大学 Radiation source direct positioning method combining arrival time difference and arrival angle information under sparse Bayesian framework

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107703477A (en) * 2017-09-11 2018-02-16 电子科技大学 The steady broadband array signal Wave arrival direction estimating method of standard based on block management loading
US20200057163A1 (en) * 2018-08-16 2020-02-20 Movano Inc. Bayesian Geolocation and Parameter Estimation by Retaining Channel and State Information
CN111007457A (en) * 2018-10-08 2020-04-14 哈尔滨工业大学 Radiation source direct positioning method based on block sparse Bayesian model
CN111007458A (en) * 2018-10-08 2020-04-14 哈尔滨工业大学 Radiation source direct positioning method combining arrival time difference and arrival angle information under sparse Bayesian framework

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
ZHILIN ZHANG等: "Sparse Signal Recovery With Temporally Correlated Source Vectors Using Sparse Bayesian Learning", 《IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING》, vol. 5, no. 5, 30 September 2011 (2011-09-30), pages 2 - 3, XP011373616, DOI: 10.1109/JSTSP.2011.2159773 *
陈敏求: "基于空间谱估计的辐射源直接定位技术", 《中国博士学位论文全文数据库 信息科技辑》, 15 January 2021 (2021-01-15), pages 1 - 4 *

Similar Documents

Publication Publication Date Title
CN109655799B (en) IAA-based covariance matrix vectorization non-uniform sparse array direction finding method
CN109490819B (en) Sparse Bayesian learning-based method for estimating direction of arrival of wave in a lattice
CN111337893B (en) Off-grid DOA estimation method based on real-value sparse Bayesian learning
CN104977558A (en) Distributed source center direction-of-arrival estimation method based on Bayesian compressed perception
CN111707985A (en) Off-grid DOA estimation method based on covariance matrix reconstruction
CN110275158A (en) Wideband radar echo-signal method for parameter estimation based on Bayes's compressed sensing
CN106501765A (en) A kind of Maximum Likelihood DOA Estimation based on quadratic sum and Semidefinite Programming
CN108398659B (en) Direction-of-arrival estimation method combining matrix beam and root finding MUSIC
CN111007457B (en) Radiation source direct positioning method based on block sparse Bayesian model
CN102175990A (en) Method and device for tracking direction of arrival based on Luenberger observer and subspace updating
CN110954862B (en) Radiation source direct positioning method based on global narrow-band model under sparse Bayesian framework
CN114720938A (en) Large-scale antenna array single-bit sampling DOA estimation method based on depth expansion
US20220349986A1 (en) Radar communication with interference suppression
CN114374927A (en) Multi-target direct positioning method based on block sparse Bayes
CN110764086B (en) Bayesian radar correlation imaging method based on disturbance matrix estimation
CN109633634B (en) Sparse Bayesian learning-based MIMO radar wave separation direction and wave arrival direction joint estimation method
CN112731273B (en) Low-complexity signal direction-of-arrival estimation method based on sparse Bayesian
CN111007458B (en) Radiation source direct positioning method combining arrival time difference and arrival angle information under sparse Bayesian framework
CN115248413A (en) Off-grid signal direction-of-arrival estimation method suitable for non-uniform linear array
CN109298384B (en) Non-uniform linear array direction of arrival angle estimation method based on variational Bayes inference
CN108957386B (en) Incoherent direction of arrival estimation method under phase error
CN110764085B (en) Variational Bayes radar correlation imaging method combined with minimum mean square error estimation
CN114415109B (en) Direct positioning method for sparse Bayesian learning
CN116299193A (en) MIMO radar intelligent DOA estimation method
CN117092585B (en) Single-bit quantized DoA estimation method, system and intelligent terminal

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