CN108768543A - Self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication is empty fast - Google Patents

Self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication is empty fast Download PDF

Info

Publication number
CN108768543A
CN108768543A CN201810529231.XA CN201810529231A CN108768543A CN 108768543 A CN108768543 A CN 108768543A CN 201810529231 A CN201810529231 A CN 201810529231A CN 108768543 A CN108768543 A CN 108768543A
Authority
CN
China
Prior art keywords
interference
signal
indicate
fast
adaptive
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201810529231.XA
Other languages
Chinese (zh)
Other versions
CN108768543B (en
Inventor
王峰
周易
龚道银
陈哲
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.)
Hohai University HHU
Original Assignee
Hohai University HHU
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 Hohai University HHU filed Critical Hohai University HHU
Priority to CN201810529231.XA priority Critical patent/CN108768543B/en
Publication of CN108768543A publication Critical patent/CN108768543A/en
Application granted granted Critical
Publication of CN108768543B publication Critical patent/CN108768543B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B13/00—Transmission systems characterised by the medium used for transmission, not provided for in groups H04B3/00 - H04B11/00
    • H04B13/02—Transmission systems in which the medium consists of the earth or a large mass of water thereon, e.g. earth telegraphy
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B15/00—Suppression or limitation of noise or interference
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
    • H04L25/03006—Arrangements for removing intersymbol interference
    • H04L25/03012—Arrangements for removing intersymbol interference operating in the time domain
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/03—Shaping networks in transmitter or receiver, e.g. adaptive shaping networks
    • H04L25/03006—Arrangements for removing intersymbol interference
    • H04L25/03159—Arrangements for removing intersymbol interference operating in the frequency domain

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Power Engineering (AREA)
  • Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)

Abstract

Self-adaptive processing algorithm when empty fast the invention discloses a kind of multiple features fusion cognition type underwater sound communication introduces the interference cognitive function analyzed based on interference characteristic when traditional empty fast on the basis of adaptive array processing algorithm；Interference source number is calculated using array covariance matrix feature decomposition, and target user direction and interference arrival bearing are estimated using MUSIC spatial spectrum analysis；And interference signal separation is carried out using the blind source separation algorithm based on the maximum Fast-ICA of negentropy, temporal signatures are analyzed using envelope detected, time-frequency spectrum is analyzed using Short Time Fourier Transform, interference type is identified to extract interference characteristic；Adaptive cancellation when suitable interference sample carries out empty fast is finally chosen according to interference type, improves the convergence rate of adaptive algorithm.Advantageous effect：The dimension-reduction treatment of adaptive algorithm and offset sample selection when realizing empty fast, ensure that it is empty soon when self-adaptive processing fast and reliable convergence.

Description

Self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication is empty fast
Technical field
Self-adaptive processing algorithm when empty fast the present invention relates to a kind of underwater sound communication, recognizes more particularly to a kind of multiple features fusion Know self-adaptive processing algorithm when type underwater sound communication is empty fast, belongs to underwater sound communication Anti-Jamming Technique field.
Background technology
Since electromagnetic wave is decayed seriously in water, limited transmission distance, water sound communication technique become ocean development explore and The important means of submarine communication.Shallow-sea underwater acoustic communication is faced with from various interference, underwater sound communication common interference form master To be niose-modulating-frenquency jamming, impulse disturbances and linear frequency sweep interference etc..Niose-modulating-frenquency jamming has wide bandwidth, power spectrum uniform The big feature with jamming power；Impulse disturbances have the duration short, and instantaneous power is big, and signal may because of the influence of big instantaneous power Appearance is saturated or the characteristics of amplitude limit；Linear frequency sweep interference has the characteristics that frequecy characteristic is linear.
Other users relative to destinations traffic interfere, course line ship and continuous wave CO_2 laser, marine organisms caused by storm Pipe impulse disturbances caused by with construction of the mankind in ocean, greatly reduces underwater sound communication system performance.Although big portion Point interference can by it is empty fast when self-adaptive processing algorithm effectively inhibit, but when some interference cannot directly use empty fast, is adaptive Processing Algorithm is handled, such as impulse disturbances, and adaptive algorithm convergence position is occurred by impulse disturbances to be influenced, and inhibits difficult It spends larger.
Invention content
It is a primary object of the present invention to overcome deficiency in the prior art, provide a kind of multiple features fusion cognition type water Self-adaptive processing algorithm when sound communication is empty fast realizes the dimension-reduction treatment of adaptive algorithm when sky is fast and offsets sample selection, ensures The fast and reliable convergence of self-adaptive processing when empty fast.
In order to achieve the above object, the technical solution adopted in the present invention is：
Self-adaptive processing algorithm when a kind of multiple features fusion cognition type underwater sound communication is empty fast, includes the following steps：
1) signal for receiving receiving array forms wave beam in target user direction and is exported as main channel signal, with And the array data extracted in receiving array is exported as auxiliary channel signal to interference cognition unit and disturbance ecology unit；
2) interference cognition unit calculates interference source number using array covariance matrix feature decomposition, and uses MUSIC Spatial spectrum analysis estimates target user direction and interference arrival bearing；
3) according to interference source number, the fast time-domain two dimension of blind source separating port number and spatial domain in blind source separation algorithm is determined Adaptive channel dimension in adaptive algorithm；The blind source separating port number and adaptive channel dimension are equal to interference source Number；
4) disturbance ecology unit carries out interference signal point using the blind source separation algorithm based on the maximum Fast-ICA of negentropy From to the interference signal after separation using time frequency analysis extraction time-frequency spectrum and using envelope statistics characteristic analysis extraction temporal envelope Feature, and the time-frequency spectrum of extraction and temporal envelope Fusion Features are perceived as into interference multiple features；
5) according to interference multiple features, disturbance ecology unit carries out interference type classification using decision tree classifier；According to point The good interference type of class, disturbance ecology unit selection interference sample is exported inhibits unit to interference cancellation；The interference type packet Include impulse disturbances, linear frequency sweep interference and niose-modulating-frenquency jamming；
6) adaptive cancellation when interference cancellation inhibits unit to carry out empty fast using the fast time-domain two-dimensional adaptive algorithm in spatial domain, To being exported after wave beam progress AF panel.
The present invention is further arranged to：Interference is calculated using array covariance matrix feature decomposition in the step 2) Source number, specifically,
The accessory channel number of extraction 2-1) is set as NAA, the sampled data of each accessory channel is Ns, then by the k moment Secondary channel data matrix is write as vector form,
Wherein, xi(k)=[xi(k),xi(k-1),…xi(k-Ns+1)]TFor i-th of secondary channel data matrix of k moment, [·]TIndicate transposition operation；
2-2) calculate the covariance matrix R of accessory channel0,
R0=x (k) xH(k)；
Wherein, x (k) is the secondary channel data matrix at k moment, []HIndicate conjugate transposition operation；
Eigenvalues Decomposition 2-3) is carried out to covariance matrix,
λi=eig [R0], i=1,2 ..., NA；
Wherein, eig [] indicates characteristic root analytic function, λiIt is expressed as ith feature root；
2-4) obtain smallest real eigenvalueMin [] indicates that minimum value finds operation；
The ratio for calculating each characteristic root and smallest real eigenvalue is denoted as ratio vector Λ,
2-5) define ratio threshold ζΛ, count ratio in ratio vector Λ and be more than ratio threshold ζΛNumber, it is as dry Disturb source number.
The present invention is further arranged to：Target user side is estimated using MUSIC spatial spectrum analysis in the step 2) To with interference arrival bearing, specifically,
Using the orthogonality of signal subspace and noise subspace, construction space spectral function is detected by spectrum peak search The arrival bearing, space spectral function is interfered to be in underwater sound communication
Wherein, θnN-th of discrete angular spatial noise, U when being scanned for spatial beamsNFor spatial noise, sweared by noise characteristic Amount is constituted, a (θn) it is signal section space, it is made of character vector of signals, []HIndicate conjugate transposition operation.
The present invention is further arranged to：Blind source separating of the use based on the maximum Fast-ICA of negentropy in the step 4) Algorithm carries out interference signal separation, specifically,
4-1-A) the observation signal matrix X of input is pre-processed, pretreatment includes going mean value and albefaction, after pretreatment Obtain prewhitening signal Z；
4-1-B) by array covariance matrix feature decomposition known disturbances source number N, random initializtion weight vector Wp, In, WpSubscript p indicate iterations p times；
Weights 4-1-C) are updated,
Wherein, mathematic expectaion is sought in E { } expressions, []TFor transposition operation, g'() it indicates to seek first derivative；
NoteThen
Coefficient a1∈ [1,2], i.e. value are 1 or 2, log2[] indicates that the logarithm bottom of for 2, cosh () indicate hyperbolic Cosine；
4-1-D) each iterative extraction WpWhen, decorrelative transformation, i.e. orthogonalization projection operation are carried out, operation formula is
Wherein, WjValue for the weight vector of iteration j, j is [1, p-1]；
4-1-E) normalized, formula are
Wherein, | | | | it indicates to calculate two norm of Euclid；
If 4-1-F) algorithm is not restrained, return to step 4-1-C) it continues to execute；
If algorithmic statement, an independent element y is found outi=WpX。
The present invention is further arranged to：Time-frequency spectrum is extracted using time frequency analysis in the step 4), specifically,
Short Time Fourier Transform 4-2-A) is carried out to the interference signal y (k) after separation, transforming function transformation function is,
Wherein, τ is the hop count of Short Time Fourier Transform, and k is discrete time, NsFor the points of accumulation, h (k) is window function, [·]*Indicate conjugate operation；
Frequency peak search 4-2-B) is carried out to Short Time Fourier Transform result, searching for formula is,
[Pmax(τ),Fmax(τ)]=max (Ω (τ,:))；
Wherein, max () indicates that maximum value finds operation；
4-2-C) calculate the first-order difference υ of frequency peak；
4-2-D) define difference variance thresholding ζF, when the first-order difference υ of frequency peak is less than difference variance thresholding ζF, then sentence Disconnected interference signal is interfered for linear frequency sweep；When the first-order difference υ of frequency peak is more than difference variance thresholding ζF, then judge interference letter Number be impulse disturbances or niose-modulating-frenquency jamming.
The present invention is further arranged to：It is special using envelope statistics characteristic analysis extraction temporal envelope in the step 4) Sign, specifically,
Pulsewidth feature 4-3-A) is extracted to the interference signal after separation, judges that interference signal is continuous wave CO_2 laser or pulse Interference；
According to the rising edge and failing edge of envelope extraction pulse, the sampling number of rising edge and failing edge is denoted as N respectivelydiWith Nri；
According to the difference and sample rate f of the rising edge of i-th of pulse and failing edgesCalculate interference signal pulse width Bi, Calculation formula is Bi=(Ndi-Nri)/fs；
Define pulsewidth thresholding ζB, when interference signal pulse width is less than pulsewidth thresholding ζB, then judge that interference signal is believed for pulse Number；When interference signal pulse width is more than pulsewidth thresholding ζB, then judge that interference signal is interfered for continuous wave CO_2 laser, i.e. linear frequency sweep Or niose-modulating-frenquency jamming；
Duty ratio feature 4-3-B) is extracted to the interference signal after separation, judges that interference signal is continuous wave CO_2 laser or arteries and veins Punching interference；
Duty ratio refers to that the ratio shared by pulse, calculation formula are in entire sample,
Wherein, NcIt counts for specimen sample, NBFor according to the pulse number of rising edge and failing edge calculating；
Define duty ratio thresholding ζK, when interference signal duty ratio is more than duty ratio thresholding ζK, then judge that interference signal is continuous Wave interference, i.e. linear frequency sweep interference or niose-modulating-frenquency jamming；When interference signal duty ratio is less than duty ratio thresholding ζK, then judge dry It is impulse disturbances to disturb signal.
The present invention is further arranged to：Interference type classification, tool are carried out using decision tree classifier in the step 5) Body is,
5-1) define the category set C={ y of interference type classification1,y2,y3}；
Wherein, y1Indicate impulse disturbances, y2Indicate linear frequency sweep interference, y3Indicate niose-modulating-frenquency jamming；
Determine interference characteristic attribute and division；
If H={ a1,a2,a3It is characterized attribute set, characteristic attribute a1Indicate the pulsewidth in temporal envelope feature, a2It indicates Duty ratio in temporal envelope feature, a3Indicate the first-order difference of the frequency peak of time-frequency spectrum；
5-2) obtain interference data training sample；
The training sample set with 1000 different parameters is built, impulse disturbances account for 40%, and linear frequency sweep interferes and noise Frequency modulation interference respectively accounts for 30%；
5-3) calculate the information gain of each interference characteristic；
Calculation formula is
Wherein, m indicates interference type number, piIndicate the probability that i-th of interference type occurs；
Calculate unfiled preceding entropy D0,
D0=-0.4*log20.4-0.3*log20.3-0.3*log20.3=1.57；
It calculates separately by characteristic attribute a1The entropy D of classification1, by a2The entropy D of classification2, by a3The entropy D of classification3,
Calculate separately D1、D2And D3With D0Difference, respectively 0.97,0.97 and 0.88；
5-4) generate decision tree；
Characteristic attribute a is selected according to maximum gain2As the first classification of disturbance node, by a2Sample is divided into 2 set, Since the set less than duty ratio thresholding is entirely impulse disturbances y1, then next just for the set choosing more than duty ratio thresholding Select class node；
Select characteristic attribute a3As the second classification of disturbance node, by a3The sample that will be greater than duty ratio thresholding is divided into 2 collection It closes, since the set less than difference variance thresholding is entirely that linear frequency sweep interferes y2, it is whole more than the set of difference variance thresholding It is niose-modulating-frenquency jamming y3, set can be completely separable, and node split terminates；
5-5) decision tree classifier carries out interference type classification by judgement；
If the duty ratio of interference signal is less than duty ratio thresholding, judgement interference signal is impulse disturbances；Otherwise judge interference Whether the first-order difference of the frequency peak of signal is less than difference variance thresholding, and if judging less than if, interference signal is dry for linear frequency sweep It disturbs, otherwise judges that interference signal is niose-modulating-frenquency jamming.
The present invention is further arranged to：Being carried out using the fast time-domain two-dimensional adaptive algorithm in spatial domain in the step 6) is empty Adaptive cancellation when fast, specifically,
The secondary channel data matrix at known k moment uses one group of length for NwWeight coefficient indicate adaptive-filtering Each FIR filter of device group is write as filter group vector matrix
Wherein, Indicate NwI-th of delay cell of a array element adds Weights；
Based on linearly constrained minimum variance, i.e. LCMV criterion, optimum right vector is solved by following formula,
wopt=Rxx -1ψ,
Rxx=E { x (k) xH(k) },
ψ==E { x (k) sH(k-Δ)}；
Wherein, woptFor optimum right vector, RxxFor sky when covariance matrix, ψ is cross correlation vector, and mathematics is sought in E { } expressions It is expected that []HIndicate conjugate transposition operation, []-1Indicate that inversion operation, s (k) indicate that main beam signal, Δ indicate main beam Signal time delay.
Compared with prior art, the invention has the advantages that：
Self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication provided by the invention is empty fast, when traditional empty fast The interference cognitive function analyzed based on interference characteristic is introduced on the basis of adaptive array processing algorithm；By interfering cognition single Member calculates interference source number using array covariance matrix feature decomposition, and estimates target using MUSIC spatial spectrum analysis User direction and interference arrival bearing.Interference number is substantially carried out the application of two aspects：It is calculated first, auxiliary chooses blind source separating Separation dimension in method reduces the complexity of blind source separation algorithm to effectively reduce the quantity of multichannel；Second is that auxiliary is chosen Spatial domain dimension when empty fast in adaptive algorithm, carries out algorithm dimension-reduction treatment, self-adaptive processing algorithm when to reduce empty fast Calculation amount.And interference letter is carried out using the blind source separation algorithm based on the maximum Fast-ICA of negentropy by disturbance ecology unit Number separation, using envelope detected analyze temporal signatures, using Short Time Fourier Transform analyze time-frequency spectrum, to extract interference characteristic Identify interference type.Adaptive cancellation when suitable interference sample carries out empty fast is finally chosen according to interference type, is improved adaptive Answer convergence speed of the algorithm.It the dimension-reduction treatment of adaptive algorithm and offsets sample when the present invention realizes empty fast and chooses, ensure that The fast and reliable convergence of self-adaptive processing when empty fast.
The above is only the general introduction of technical solution of the present invention, in order to be better understood upon the technological means of the present invention, under In conjunction with attached drawing, the invention will be further described in face.
Description of the drawings
Fig. 1 is the functional block diagram of self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication of the present invention is empty fast；
Fig. 2 is the process chart of self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication of the present invention is empty fast；
Fig. 3 is to extract time-frequency spectrum in self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication of the present invention is empty fast Functional block diagram；
Fig. 4 is to carry out interference type in self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication of the present invention is empty fast The functional block diagram of classification；
Fig. 5 is the decision tree generated in self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication of the present invention is empty fast Structural schematic diagram；
Fig. 6 is the MUSIC spatial spectrum direction-finding charts of embodiment；
Fig. 7 be embodiment blind source separating after interfere time-domain diagram；
Fig. 8 is interference time-frequency figure after the blind source separating of embodiment；
Fig. 9 is the classification of disturbance accuracy under the different signal-to-noise ratio of embodiment；
Figure 10 is power diagram before and after the AF panel of embodiment；
Figure 11 is that the front and back correlation that offsets of embodiment is schemed.
Specific implementation mode
With reference to the accompanying drawings of the specification, the present invention is further illustrated.
Self-adaptive processing algorithm when a kind of multiple features fusion cognition type underwater sound communication of invention offer is empty fast, such as Fig. 1 and Fig. 2 institutes Show, originally includes the following steps：
1) signal for receiving receiving array forms wave beam in target user direction and is exported as main channel signal, with And the array data extracted in receiving array is exported as auxiliary channel signal to interference cognition unit and disturbance ecology unit.
2) interference cognition unit calculates interference source number using array covariance matrix feature decomposition, and uses MUSIC Spatial spectrum analysis estimates target user direction and interference arrival bearing.
Interference source number is calculated using array covariance matrix feature decomposition in the step 2), specifically,
The accessory channel number of extraction 2-1) is set as NAA, the sampled data of each accessory channel is Ns, then by the k moment Secondary channel data matrix is write as vector form,
Wherein, xi(k)=[xi(k),xi(k-1),…xi(k-Ns+1)]TFor i-th of secondary channel data matrix of k moment, [·]TIndicate transposition operation；
2-2) calculate the covariance matrix R of accessory channel0,
R0=x (k) xH(k)；
Wherein, x (k) is the secondary channel data matrix at k moment, []HIndicate conjugate transposition operation；
Eigenvalues Decomposition 2-3) is carried out to covariance matrix,
λi=eig [R0], i=1,2 ..., NA；
Wherein, eig [] indicates characteristic root analytic function, λiIt is expressed as ith feature root；
2-4) obtain smallest real eigenvalueMin [] indicates that minimum value finds operation；
The ratio for calculating each characteristic root and smallest real eigenvalue is denoted as ratio vector Λ,
2-5) define ratio threshold ζΛ, count ratio in ratio vector Λ and be more than ratio threshold ζΛNumber, it is as dry Disturb source number.
Target user direction and interference arrival bearing, tool are estimated using MUSIC spatial spectrum analysis in the step 2) Body is,
Using the orthogonality of signal subspace and noise subspace, construction space spectral function is detected by spectrum peak search The arrival bearing, space spectral function is interfered to be in underwater sound communication
Wherein, θnN-th of discrete angular spatial noise, U when being scanned for spatial beamsNFor spatial noise, sweared by noise characteristic Amount is constituted, a (θn) it is signal section space, it is made of character vector of signals, []HIndicate conjugate transposition operation.
3) according to interference source number, the fast time-domain two dimension of blind source separating port number and spatial domain in blind source separation algorithm is determined Adaptive channel dimension in adaptive algorithm；The blind source separating port number and adaptive channel dimension are equal to interference source Number.
4) disturbance ecology unit carries out interference signal point using the blind source separation algorithm based on the maximum Fast-ICA of negentropy From to the interference signal after separation using time frequency analysis extraction time-frequency spectrum and using envelope statistics characteristic analysis extraction temporal envelope Feature, and the time-frequency spectrum of extraction and temporal envelope Fusion Features are perceived as into interference multiple features.
Blind source separation algorithm of the use based on the maximum Fast-ICA of negentropy in the step 4) carries out interference signal point From, specifically,
4-1-A) the observation signal matrix X of input is pre-processed, pretreatment includes going mean value and albefaction, after pretreatment Obtain prewhitening signal Z；
4-1-B) by array covariance matrix feature decomposition known disturbances source number N, random initializtion weight vector Wp, In, WpSubscript p indicate iterations p times；
Weights 4-1-C) are updated,
Wherein, mathematic expectaion is sought in E { } expressions, []TFor transposition operation, g'() it indicates to seek first derivative；
NoteThen
Coefficient a1∈ [1,2], i.e. value are 1 or 2, log2[] indicates that the logarithm bottom of for 2, cosh () indicate hyperbolic Cosine；
4-1-D) each iterative extraction WpWhen, decorrelative transformation, i.e. orthogonalization projection operation are carried out, operation formula is
Wherein, WjValue for the weight vector of iteration j, j is [1, p-1]；
4-1-E) normalized, formula are
Wherein, | | | | it indicates to calculate two norm of Euclid；
If 4-1-F) algorithm is not restrained, return to step 4-1-C) it continues to execute；
If algorithmic statement, an independent element y is found outi=WpX。
Time-frequency spectrum is extracted using time frequency analysis in the step 4), specifically, as shown in figure 3,
Short Time Fourier Transform 4-2-A) is carried out to the interference signal y (k) after separation, transforming function transformation function is,
Wherein, τ is the hop count of Short Time Fourier Transform, and k is discrete time, NsFor the points of accumulation, h (k) is window function, [·]*Indicate conjugate operation；
Frequency peak search 4-2-B) is carried out to Short Time Fourier Transform result, searching for formula is,
[Pmax(τ),Fmax(τ)]=max (Ω (τ,:))；
Wherein, max () indicates that maximum value finds operation；
4-2-C) calculate the first-order difference υ of frequency peak；
4-2-D) define difference variance thresholding ζF, when the first-order difference υ of frequency peak is less than difference variance thresholding ζF, then sentence Disconnected interference signal is interfered for linear frequency sweep；When the first-order difference υ of frequency peak is more than difference variance thresholding ζF, then judge interference letter Number be impulse disturbances or niose-modulating-frenquency jamming.
Temporal envelope feature is extracted using envelope statistics characteristic analysis in the step 4), specifically,
Pulsewidth feature 4-3-A) is extracted to the interference signal after separation, judges that interference signal is continuous wave CO_2 laser or pulse Interference；
According to the rising edge and failing edge of envelope extraction pulse, the sampling number of rising edge and failing edge is denoted as N respectivelydiWith Nri；
According to the difference and sample rate f of the rising edge of i-th of pulse and failing edgesCalculate interference signal pulse width Bi, Calculation formula is Bi=(Ndi-Nri)/fs；
Define pulsewidth thresholding ζB, when interference signal pulse width is less than pulsewidth thresholding ζB, then judge that interference signal is believed for pulse Number；When interference signal pulse width is more than pulsewidth thresholding ζB, then judge that interference signal is interfered for continuous wave CO_2 laser, i.e. linear frequency sweep Or niose-modulating-frenquency jamming；
Duty ratio feature 4-3-B) is extracted to the interference signal after separation, judges that interference signal is continuous wave CO_2 laser or arteries and veins Punching interference；
Duty ratio refers to that the ratio shared by pulse, calculation formula are in entire sample,
Wherein, NcIt counts for specimen sample, NBFor according to the pulse number of rising edge and failing edge calculating；
Define duty ratio thresholding ζK, when interference signal duty ratio is more than duty ratio thresholding ζK, then judge that interference signal is continuous Wave interference, i.e. linear frequency sweep interference or niose-modulating-frenquency jamming；When interference signal duty ratio is less than duty ratio thresholding ζK, then judge dry It is impulse disturbances to disturb signal.
In conclusion the time-frequency spectrum of extraction and temporal envelope Fusion Features are perceived as interference multiple features, multiple features are interfered The results are shown in Table 1 for fusion cognition.
Table 1
For different interference signals, the pulsewidth of impulse disturbances is less than ζB, duty ratio is less than ζK, a scale of frequency peak Divide and is more than ζF；The pulsewidth of linear frequency sweep interference is more than ζB, duty ratio is more than ζK, the first-order difference of frequency peak is less than ζF；Noise tune The pulsewidth of frequency interference is more than ζB, duty ratio is more than ζK, the first-order difference of frequency peak is more than ζF。
5) according to interference multiple features, disturbance ecology unit carries out interference type classification using decision tree classifier；According to point The good interference type of class, disturbance ecology unit selection interference sample is exported inhibits unit to interference cancellation；The interference type packet Include impulse disturbances, linear frequency sweep interference and niose-modulating-frenquency jamming.
Decision tree divides the principle that interference collects：Compare the comentropy by the sorted interference data of different characteristic, by certain Comentropy after one tagsort is reduced most, then this feature is optimal classification feature.
Interference type classification is carried out using decision tree classifier in the step 5), specifically, as shown in figure 4,
5-1) define the category set C={ y of interference type classification1,y2,y3}；
Wherein, y1Indicate impulse disturbances, y2Indicate linear frequency sweep interference, y3Indicate niose-modulating-frenquency jamming；
Determine interference characteristic attribute and division；
If H={ a1,a2,a3It is characterized attribute set, characteristic attribute a1Indicate the pulsewidth in temporal envelope feature, a2It indicates Duty ratio in temporal envelope feature, a3Indicate the first-order difference of the frequency peak of time-frequency spectrum；
5-2) obtain interference data training sample；
The training sample set with 1000 different parameters is built, impulse disturbances account for 40%, and linear frequency sweep interferes and noise Frequency modulation interference respectively accounts for 30%；
5-3) calculate the information gain of each interference characteristic；
Calculation formula is
Wherein, m indicates interference type number, piIndicate the probability that i-th of interference type occurs；
Calculate unfiled preceding entropy D0,
D0=-0.4*log20.4-0.3*log20.3-0.3*log20.3=1.57；
It calculates separately by characteristic attribute a1The entropy D of classification1, by a2The entropy D of classification2, by a3The entropy D of classification3,
Calculate separately D1、D2And D3With D0Difference, respectively 0.97,0.97 and 0.88；
Decision tree 5-4) is generated, as shown in Figure 5；
Characteristic attribute a is selected according to maximum gain2As the first classification of disturbance node, by a2Sample is divided into 2 set, Since the set less than duty ratio thresholding is entirely impulse disturbances y1, then next just for the set choosing more than duty ratio thresholding Select class node；
Select characteristic attribute a3As the second classification of disturbance node, by a3The sample that will be greater than duty ratio thresholding is divided into 2 collection It closes, since the set less than difference variance thresholding is entirely that linear frequency sweep interferes y2, it is whole more than the set of difference variance thresholding It is niose-modulating-frenquency jamming y3, set can be completely separable, and node split terminates；
5-5) decision tree classifier carries out interference type classification by judgement；
If the duty ratio of interference signal is less than duty ratio thresholding, judgement interference signal is impulse disturbances；Otherwise judge interference Whether the first-order difference of the frequency peak of signal is less than difference variance thresholding, and if judging less than if, interference signal is dry for linear frequency sweep It disturbs, otherwise judges that interference signal is niose-modulating-frenquency jamming.
6) adaptive cancellation when interference cancellation inhibits unit to carry out empty fast using the fast time-domain two-dimensional adaptive algorithm in spatial domain, To being exported after wave beam progress AF panel.
Adaptive cancellation when carrying out empty fast using the fast time-domain two-dimensional adaptive algorithm in spatial domain in the step 6), specifically For,
The secondary channel data matrix at known k moment uses one group of length for NwWeight coefficient indicate adaptive-filtering Each FIR filter of device group is write as filter group vector matrix
Wherein, Indicate NwI-th of delay cell of a array element adds Weights；
Based on linearly constrained minimum variance, i.e. LCMV criterion, optimum right vector is solved by following formula,
wopt=Rxx -1ψ,
Rxx=E { x (k) xH(k) },
ψ==E { x (k) sH(k-Δ)}；
Wherein, woptFor optimum right vector, RxxFor sky when covariance matrix, ψ is cross correlation vector, and mathematics is sought in E { } expressions It is expected that []HIndicate conjugate transposition operation, []-1Indicate that inversion operation, s (k) indicate that main beam signal, Δ indicate main beam Signal time delay.
Embodiment
Self-adaptive processing algorithm when verifying multiple features fusion cognition type underwater sound communication sky of the present invention soon by Computer Simulation Identification to interference and rejection.One subscriber signal of setting and three kinds of interference signals, subscriber signal are in MATLAB emulation Two-phase PSK (BPSK) is modulated, and three kinds of interference are respectively impulse disturbances, niose-modulating-frenquency jamming and linear frequency sweep interference.It receives It is half-wavelength that array, which uses the linear array of Unit 16, cell spacing, and array is formed wave beam and weighted using the Chebyshev of 40dB, received Signal is 4000 sampled points per frame length, and user's incidence angle is 0 ° (known), and interference incidence angle is respectively -5 °, 5 ° and 10 °, It is dry in channel to make an uproar than for 30dB, signal-to-noise ratio is -20dB, jamming-to-signal ratio 50dB.Interfere the thresholding setting of multiple features as shown in table 2.
Table 2
Emulation 1：Interfere cognitive simulation
After array forms wave beam in target user direction, first by array covariance matrix feature decomposition, through formula It is 3 that notable feature root number, which is calculated, i.e., interference source number is 3.User and interference are estimated by MUSIC spatial spectrum analysis again Arrival bearing, as shown in fig. 6, the direction for measuring target and interference is respectively -5 °, 0 °, 5 ° and 10 °, since known 0 ° be use Family direction, the direction that can measure interference are respectively -5 °, 5 ° and 10 °.
The separation of interference signal is realized using the blind source separation algorithm based on the maximum Fast-ICA of negentropy, separating resulting is such as Shown in Fig. 7.
Envelope characteristic is extracted respectively to three kinds of interference signals after separation, the pulsewidth of disturbed one is 80 sampled points, duty ratio It is 0.22；The pulsewidth of interference 2 is 4000 sampled points, duty ratio 1；The pulsewidth of interference 3 is 4000 sampled points, and duty ratio is 1。
Time-frequency spectrum is extracted using time frequency analysis to the interference signal after separation, that treated is dry using Short Time Fourier Transform Signal time-frequency figure is disturbed, as shown in Figure 8；Fig. 8-(b) is interference 2 to interfere time-frequency figure, Fig. 8-(a) after the blind source separating of disturbed one After blind source separating interfere time-frequency figure, Fig. 8-(c) be interfere 3 blind source separating after interference time-frequency figure.Extract the frequency peak of time-frequency spectrum Value, the variance of the first-order difference of three kinds of interfering frequency peak values of calculating, respectively 4747,6723,0.1538.
The interference characteristic parameter extracted in emulation, as shown in table 3.
Table 3
According to the feature that table 3 extracts, classified using decision tree classifier.Disturbed one duty ratio is 0.22, is less than duty Than thresholding 0.7, it is judged to impulse disturbances.It is 1 to interfere 2 duty ratios, is more than duty ratio thresholding 0.7, frequency peak first-order difference Variance is 6723, is more than difference variance thresholding 1, so being finally judged to niose-modulating-frenquency jamming.The duty ratio of interference 3 is 1, is more than The variance of duty ratio thresholding 0.7, frequency peak first-order difference is 0.1538, is less than difference variance thresholding 1, so being finally judged to Linear frequency sweep interferes.Number is interfered, direction, type is consistent with simulated conditions setting, demonstrates the effective of cognitive interference algorithm Property.
Change the signal-to-noise ratio in phantom channel, flow diagram according to the invention classifies to interference, and statistical classification is just For true rate with the change curve of signal-to-noise ratio, classification of disturbance accuracy is as shown in Figure 9.As can be seen from Figure 9, signal-to-noise ratio is higher than 8dB When, which can obtain higher disturbance ecology rate, and accuracy can reach 99% or more.
Emulation 2：Self-adaptive processing emulates when empty fast
Self-adaptive processing algorithm inhibits interference when choosing sample after identification interference type, then using empty fast, such as Impulse disturbances choose the sample that impulse disturbances occur.There are three interference in this emulation, choose three accessory channels.It is done in emulation 1 It is as shown in Figure 10 to disturb the front and back figure of inhibition, cancellation ratio 16.82dB.Inhibit front and back coherence map as shown in figure 11, as seen from Figure 11, suppression The front and back correlation peak of system is identical, offsets rear secondary lobe and is depressed, it was demonstrated that after offseting the result is that subscriber signal, demonstrates inhibition and calculates The validity of method.
The innovation of the invention consists in that by interfering the cognition of number to reduce blind source separation algorithm and space-time adaptive Calculation amount in algorithm realizes niose-modulating-frenquency jamming, pulse by the multiple features fusion cognition to temporal envelope and time-frequency spectrum The Classification and Identification of interference and linear frequency sweep interference, completes the selection of interference sample, self-adaptive processing when finally ensure that empty fast Reliable conveyance.
The basic principles and main features and advantage of the present invention have been shown and described above.The technical staff of the industry should Understand, the present invention is not limited to the above embodiments, and the above embodiments and description only describe the originals of the present invention Reason, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes and improvements It all fall within the protetion scope of the claimed invention.The claimed scope of the invention is by appended claims and its equivalent circle It is fixed.

Claims (8)

  1. Self-adaptive processing algorithm when 1. a kind of multiple features fusion cognition type underwater sound communication is empty fast, which is characterized in that including following step Suddenly：
    1) signal for receiving receiving array forms wave beam in target user direction and is exported as main channel signal, and takes out The array data in receiving array is taken to be exported to interference cognition unit and disturbance ecology unit as auxiliary channel signal；
    2) interference cognition unit calculates interference source number using array covariance matrix feature decomposition, and using the spaces MUSIC Spectrum analysis estimates target user direction and interference arrival bearing；
    3) according to interference source number, determine that the fast time-domain two dimension of blind source separating port number and spatial domain in blind source separation algorithm is adaptive Answer the adaptive channel dimension in algorithm；The blind source separating port number and adaptive channel dimension are equal to interference source number；
    4) disturbance ecology unit carries out interference signal separation using the blind source separation algorithm based on the maximum Fast-ICA of negentropy, right Interference signal after separation extracts temporal envelope feature using time frequency analysis extraction time-frequency spectrum and using envelope statistics characteristic analysis, And the time-frequency spectrum of extraction and temporal envelope Fusion Features are perceived as into interference multiple features；
    5) according to interference multiple features, disturbance ecology unit carries out interference type classification using decision tree classifier；According to classifying Interference type, disturbance ecology unit selection interference sample export give interference cancellation inhibit unit；The interference type includes arteries and veins Punching interference, linear frequency sweep interference and niose-modulating-frenquency jamming；
    6) adaptive cancellation when interference cancellation inhibits unit to carry out empty fast using the fast time-domain two-dimensional adaptive algorithm in spatial domain, to It is exported after carrying out AF panel to wave beam.
  2. Self-adaptive processing algorithm when 2. multiple features fusion cognition type underwater sound communication according to claim 1 is empty fast, feature It is：Interference source number is calculated using array covariance matrix feature decomposition in the step 2), specifically,
    The accessory channel number of extraction 2-1) is set as NAA, the sampled data of each accessory channel is Ns, then by the auxiliary at k moment Channel data matrix is write as vector form,
    Wherein, xi(k)=[xi(k),xi(k-1),…xi(k-Ns+1)]TFor i-th of secondary channel data matrix of k moment, []T Indicate transposition operation；
    2-2) calculate the covariance matrix R of accessory channel0,
    R0=x (k) xH(k)；
    Wherein, x (k) is the secondary channel data matrix at k moment, []HIndicate conjugate transposition operation；
    Eigenvalues Decomposition 2-3) is carried out to covariance matrix,
    λi=eig [R0], i=1,2 ..., NA；
    Wherein, eig [] indicates characteristic root analytic function, λiIt is expressed as ith feature root；
    2-4) obtain smallest real eigenvalueMin [] indicates that minimum value finds operation；
    The ratio for calculating each characteristic root and smallest real eigenvalue is denoted as ratio vector Λ,
    2-5) define ratio threshold ζΛ, count ratio in ratio vector Λ and be more than ratio threshold ζΛNumber, as interference source Number.
  3. Self-adaptive processing algorithm when 3. multiple features fusion cognition type underwater sound communication according to claim 1 is empty fast, feature It is：Target user direction and interference arrival bearing are estimated using MUSIC spatial spectrum analysis in the step 2), specifically For,
    Using the orthogonality of signal subspace and noise subspace, construction space spectral function detects the underwater sound by spectrum peak search The arrival bearing, space spectral function is interfered to be in communication
    Wherein, θnN-th of discrete angular spatial noise, U when being scanned for spatial beamsNFor spatial noise, by noise characteristic vector structure At a (θn) it is signal section space, it is made of character vector of signals, []HIndicate conjugate transposition operation.
  4. Self-adaptive processing algorithm when 4. multiple features fusion cognition type underwater sound communication according to claim 1 is empty fast, feature It is：Blind source separation algorithm of the use based on the maximum Fast-ICA of negentropy in the step 4) carries out interference signal separation, Specifically,
    4-1-A) the observation signal matrix X of input is pre-processed, pretreatment includes going mean value and albefaction, is obtained after pretreatment Prewhitening signal Z；
    4-1-B) by array covariance matrix feature decomposition known disturbances source number N, random initializtion weight vector Wp, wherein Wp's Subscript p indicates iterations p times；
    Weights 4-1-C) are updated,
    Wherein, mathematic expectaion is sought in E { } expressions, []TFor transposition operation, g'() it indicates to seek first derivative；
    NoteThen
    Coefficient a1∈ [1,2], i.e. value are 1 or 2, log2[] indicates that the logarithm bottom of for 2, cosh () indicate hyperbolic cosine；
    4-1-D) each iterative extraction WpWhen, decorrelative transformation, i.e. orthogonalization projection operation are carried out, operation formula is
    Wherein, WjValue for the weight vector of iteration j, j is [1, p-1]；
    4-1-E) normalized, formula are
    Wherein, | | | | it indicates to calculate two norm of Euclid；
    If 4-1-F) algorithm is not restrained, return to step 4-1-C) it continues to execute；
    If algorithmic statement, an independent element y is found outi=WpX。
  5. Self-adaptive processing algorithm when 5. multiple features fusion cognition type underwater sound communication according to claim 1 is empty fast, feature It is：Time-frequency spectrum is extracted using time frequency analysis in the step 4), specifically,
    Short Time Fourier Transform 4-2-A) is carried out to the interference signal y (k) after separation, transforming function transformation function is,
    Wherein, τ is the hop count of Short Time Fourier Transform, and k is discrete time, NsFor the points of accumulation, h (k) is window function, []* Indicate conjugate operation；
    Frequency peak search 4-2-B) is carried out to Short Time Fourier Transform result, searching for formula is,
    [Pmax(τ),Fmax(τ)]=max (Ω (τ,:))；
    Wherein, max () indicates that maximum value finds operation；
    4-2-C) calculate the first-order difference υ of frequency peak；
    4-2-D) define difference variance thresholding ζF, when the first-order difference υ of frequency peak is less than difference variance thresholding ζF, then judge dry Signal is disturbed to interfere for linear frequency sweep；When the first-order difference υ of frequency peak is more than difference variance thresholding ζF, then judge that interference signal is Impulse disturbances or niose-modulating-frenquency jamming.
  6. Self-adaptive processing algorithm when 6. multiple features fusion cognition type underwater sound communication according to claim 1 is empty fast, feature It is：Temporal envelope feature is extracted using envelope statistics characteristic analysis in the step 4), specifically,
    Pulsewidth feature 4-3-A) is extracted to the interference signal after separation, judges that interference signal is that continuous wave CO_2 laser or pulse are dry It disturbs；
    According to the rising edge and failing edge of envelope extraction pulse, the sampling number of rising edge and failing edge is denoted as N respectivelydiAnd Nri；
    According to the difference and sample rate f of the rising edge of i-th of pulse and failing edgesCalculate interference signal pulse width Bi, calculate Formula is Bi=(Ndi-Nri)/fs；
    Define pulsewidth thresholding ζB, when interference signal pulse width is less than pulsewidth thresholding ζB, then judge interference signal for pulse signal； When interference signal pulse width is more than pulsewidth thresholding ζB, then judge interference signal for continuous wave CO_2 laser, i.e. linear frequency sweep interference or Niose-modulating-frenquency jamming；
    Duty ratio feature 4-3-B) is extracted to the interference signal after separation, judges that interference signal is that continuous wave CO_2 laser or pulse are dry It disturbs；
    Duty ratio refers to that the ratio shared by pulse, calculation formula are in entire sample,
    Wherein, NcIt counts for specimen sample, NBFor according to the pulse number of rising edge and failing edge calculating；
    Define duty ratio thresholding ζK, when interference signal duty ratio is more than duty ratio thresholding ζK, then judge that interference signal is dry for continuous wave Disturb, i.e. linear frequency sweep interference or niose-modulating-frenquency jamming；When interference signal duty ratio is less than duty ratio thresholding ζK, then judge interference letter Number be impulse disturbances.
  7. Self-adaptive processing algorithm when 7. multiple features fusion cognition type underwater sound communication according to claim 1 is empty fast, feature It is：Interference type classification is carried out using decision tree classifier in the step 5), specifically,
    5-1) define the category set C={ y of interference type classification1,y2,y3}；
    Wherein, y1Indicate impulse disturbances, y2Indicate linear frequency sweep interference, y3Indicate niose-modulating-frenquency jamming；
    Determine interference characteristic attribute and division；
    If H={ a1,a2,a3It is characterized attribute set, characteristic attribute a1Indicate the pulsewidth in temporal envelope feature, a2Indicate time domain Duty ratio in envelope characteristic, a3Indicate the first-order difference of the frequency peak of time-frequency spectrum；
    5-2) obtain interference data training sample；
    The training sample set with 1000 different parameters is built, impulse disturbances account for 40%, and linear frequency sweep interferes and noise FM Interference respectively accounts for 30%；
    5-3) calculate the information gain of each interference characteristic；
    Calculation formula is
    Wherein, m indicates interference type number, piIndicate the probability that i-th of interference type occurs；
    Calculate unfiled preceding entropy D0,
    D0=-0.4*log20.4-0.3*log20.3-0.3*log20.3=1.57；
    It calculates separately by characteristic attribute a1The entropy D of classification1, by a2The entropy D of classification2, by a3The entropy D of classification3,
    Calculate separately D1、D2And D3With D0Difference, respectively 0.97,0.97 and 0.88；
    5-4) generate decision tree；
    Characteristic attribute a is selected according to maximum gain2As the first classification of disturbance node, by a2Sample is divided into 2 set, due to Set less than duty ratio thresholding is entirely impulse disturbances y1, then next just for the Resource selection point more than duty ratio thresholding Class node；
    Select characteristic attribute a3As the second classification of disturbance node, by a3The sample that will be greater than duty ratio thresholding is divided into 2 set, Since the set less than difference variance thresholding is entirely that linear frequency sweep interferes y2, more than difference variance thresholding set be entirely make an uproar Tone frequency interference y3, set can be completely separable, and node split terminates；
    5-5) decision tree classifier carries out interference type classification by judgement；
    If the duty ratio of interference signal is less than duty ratio thresholding, judgement interference signal is impulse disturbances；Otherwise judge interference signal Frequency peak first-order difference whether be less than difference variance thresholding, if less than if judge interference signal for linear frequency sweep interfere, Otherwise judgement interference signal is niose-modulating-frenquency jamming.
  8. Self-adaptive processing algorithm when 8. multiple features fusion cognition type underwater sound communication according to claim 2 is empty fast, feature It is：Adaptive cancellation when carrying out empty fast using the fast time-domain two-dimensional adaptive algorithm in spatial domain in the step 6), specifically,
    The secondary channel data matrix at known k moment uses one group of length for NwWeight coefficient indicate sef-adapting filter group Each FIR filter is write as filter group vector matrix
    Wherein, Indicate NwThe weighted value of i-th of delay cell of a array element；
    Based on linearly constrained minimum variance, i.e. LCMV criterion, optimum right vector is solved by following formula,
    wopt=Rxx -1ψ,
    Rxx=E { x (k) xH(k) },
    ψ==E { x (k) sH(k-Δ)}；
    Wherein, woptFor optimum right vector, RxxFor sky when covariance matrix, ψ is cross correlation vector, and the mathematics phase is asked in E { } expressions It hopes, []HIndicate conjugate transposition operation, []-1Indicate that inversion operation, s (k) indicate that main beam signal, Δ indicate main beam letter Number time delay.
CN201810529231.XA 2018-05-29 2018-05-29 Multi-feature fusion cognitive underwater acoustic communication space-time and fast-time adaptive processing algorithm Active CN108768543B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810529231.XA CN108768543B (en) 2018-05-29 2018-05-29 Multi-feature fusion cognitive underwater acoustic communication space-time and fast-time adaptive processing algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810529231.XA CN108768543B (en) 2018-05-29 2018-05-29 Multi-feature fusion cognitive underwater acoustic communication space-time and fast-time adaptive processing algorithm

Publications (2)

Publication Number Publication Date
CN108768543A true CN108768543A (en) 2018-11-06
CN108768543B CN108768543B (en) 2021-06-08

Family

ID=64003520

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810529231.XA Active CN108768543B (en) 2018-05-29 2018-05-29 Multi-feature fusion cognitive underwater acoustic communication space-time and fast-time adaptive processing algorithm

Country Status (1)

Country Link
CN (1) CN108768543B (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109412640A (en) * 2018-11-22 2019-03-01 东南大学 A kind of non-linear, digital self-interference cancellation element and method applied to full-duplex communication transceiver
CN110464345A (en) * 2019-08-22 2019-11-19 北京航空航天大学 A kind of separate head bioelectrical power signal interference elimination method and system
CN110716171A (en) * 2019-08-28 2020-01-21 上海无线电设备研究所 Polarization DOA joint estimation method based on genetic algorithm
CN111307182A (en) * 2020-03-06 2020-06-19 宁波飞芯电子科技有限公司 Data processing method and array type sensor
CN111796261A (en) * 2020-06-12 2020-10-20 中国船舶重工集团公司第七二四研究所 Radar signal self-adaptive detection method based on frequency domain multi-channel statistics
CN113067646A (en) * 2021-03-30 2021-07-02 哈尔滨工程大学 Full duplex underwater acoustic communication machine for single carrier communication
CN116540064A (en) * 2023-04-07 2023-08-04 亚之捷智能装备(江苏)有限公司 Interference monitoring method and system of signal processing circuit based on big data

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7684769B2 (en) * 2005-12-16 2010-03-23 Korea Advanced Institute Of Science And Technology Data communication apparatus and module using human body
EP1891457B1 (en) * 2005-06-13 2011-03-23 WFS Technologies Limited Underwater navigation
CN101997616A (en) * 2010-10-25 2011-03-30 中国船舶重工集团公司第七一五研究所 Vector array MIMO-based high-speed underwater sound communication method
CN102333052A (en) * 2011-05-26 2012-01-25 西北工业大学 Underwater sound signal blind deconvolution method suitable for shallow sea low-frequency conditions
CN104469836A (en) * 2014-11-24 2015-03-25 河海大学常州校区 Method for building multi-dimension trust model in underwater sensor network
CN105553894A (en) * 2016-01-19 2016-05-04 南京航空航天大学 Blind source separation method based on parallel factor compressed sensing in sound vector array
CN106937348A (en) * 2017-02-21 2017-07-07 青岛科技大学 Underwater wireless Sensor Network network-building method and underwater wireless sensing net node device

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1891457B1 (en) * 2005-06-13 2011-03-23 WFS Technologies Limited Underwater navigation
US7684769B2 (en) * 2005-12-16 2010-03-23 Korea Advanced Institute Of Science And Technology Data communication apparatus and module using human body
CN101997616A (en) * 2010-10-25 2011-03-30 中国船舶重工集团公司第七一五研究所 Vector array MIMO-based high-speed underwater sound communication method
CN102333052A (en) * 2011-05-26 2012-01-25 西北工业大学 Underwater sound signal blind deconvolution method suitable for shallow sea low-frequency conditions
CN104469836A (en) * 2014-11-24 2015-03-25 河海大学常州校区 Method for building multi-dimension trust model in underwater sensor network
CN105553894A (en) * 2016-01-19 2016-05-04 南京航空航天大学 Blind source separation method based on parallel factor compressed sensing in sound vector array
CN106937348A (en) * 2017-02-21 2017-07-07 青岛科技大学 Underwater wireless Sensor Network network-building method and underwater wireless sensing net node device

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
王峰: "认知型宽带恒模阵列水声通信多用户信号分离与干扰抑制算法", 《数据采集及处理》 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109412640A (en) * 2018-11-22 2019-03-01 东南大学 A kind of non-linear, digital self-interference cancellation element and method applied to full-duplex communication transceiver
CN110464345A (en) * 2019-08-22 2019-11-19 北京航空航天大学 A kind of separate head bioelectrical power signal interference elimination method and system
CN110716171A (en) * 2019-08-28 2020-01-21 上海无线电设备研究所 Polarization DOA joint estimation method based on genetic algorithm
CN111307182A (en) * 2020-03-06 2020-06-19 宁波飞芯电子科技有限公司 Data processing method and array type sensor
CN111796261A (en) * 2020-06-12 2020-10-20 中国船舶重工集团公司第七二四研究所 Radar signal self-adaptive detection method based on frequency domain multi-channel statistics
CN111796261B (en) * 2020-06-12 2024-05-24 中国船舶集团有限公司第七二四研究所 Radar signal self-adaptive detection method based on frequency domain multichannel statistics
CN113067646A (en) * 2021-03-30 2021-07-02 哈尔滨工程大学 Full duplex underwater acoustic communication machine for single carrier communication
CN116540064A (en) * 2023-04-07 2023-08-04 亚之捷智能装备(江苏)有限公司 Interference monitoring method and system of signal processing circuit based on big data
CN116540064B (en) * 2023-04-07 2024-02-02 亚之捷智能装备(江苏)有限公司 Interference monitoring method and system of signal processing circuit based on big data

Also Published As

Publication number Publication date
CN108768543B (en) 2021-06-08

Similar Documents

Publication Publication Date Title
CN108768543A (en) Self-adaptive processing algorithm when multiple features fusion cognition type underwater sound communication is empty fast
CN108680910B (en) Frequency modulation broadcast external radiation source radar target detection method based on waveform cognition
CN106504763A (en) Based on blind source separating and the microphone array multiple target sound enhancement method of spectrum-subtraction
CN108828566B (en) Underwater pulse signal identification method based on towed linear array
US20020042685A1 (en) Optimal ratio estimator for multisensor systems
CN103854660B (en) A kind of four Mike's sound enhancement methods based on independent component analysis
CN105785330B (en) A kind of cognition type secondary lobe disturbance restraining method
CN112183225B (en) Underwater target signal feature extraction method based on probability latent semantic analysis
Balan et al. Statistical properties of STFT ratios for two channel systems and applications to blind source separation
CN113835068B (en) Blind source separation real-time main lobe interference resistance method based on independent component analysis
CN109884591A (en) A kind of multi-rotor unmanned aerial vehicle acoustical signal Enhancement Method based on microphone array
CN106483506B (en) A kind of cognition type self-adapting interference suppression method
CN116866129A (en) Wireless communication signal detection method
CN106405543B (en) A kind of cognition type blind source separating radiation source extracting method and its evaluation method
CN111007487B (en) Multi-base radar target detection method based on time reversal
US4654835A (en) Adaptive predictor of surface reverberation in a bistatic sonar
CN112394324A (en) Microphone array-based remote sound source positioning method and system
CN108896963B (en) Airborne radar space-time self-adaptive dimension reduction processing method
CN113805139A (en) Broadband signal sparse representation direction-of-arrival estimation method based on focusing transformation
CN107635181B (en) Multi-address sensing source feedback optimization method based on channel learning
CN110673118A (en) Active sonar single-frequency pulse train waveform design and detection algorithm
CN115932808A (en) Passive sonar intelligent detection method based on multi-feature fusion
CN113109760B (en) Multi-line spectrum combined DOA estimation and clustering method and system based on group sparsity
Vasylyshyn et al. Speech enhancement using modified SSA
CN114035157B (en) Sub-band delay estimation method and system based on expectation maximization algorithm

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