CN101478374B - Physical layer network code processing method - Google Patents
Physical layer network code processing method Download PDFInfo
- Publication number
- CN101478374B CN101478374B CN 200910077201 CN200910077201A CN101478374B CN 101478374 B CN101478374 B CN 101478374B CN 200910077201 CN200910077201 CN 200910077201 CN 200910077201 A CN200910077201 A CN 200910077201A CN 101478374 B CN101478374 B CN 101478374B
- Authority
- CN
- China
- Prior art keywords
- signal
- matrix
- mixed
- characteristic
- observation
- 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.)
- Expired - Fee Related
Links
Images
Abstract
The invention discloses an encoding operation method for physical-layer networks. The method comprises the following steps: a first source node broadcasts a first signal and a second source node broadcasts a second signal; a relay node receives a relay mixed signal, wherein the mixed signal is the signal that reaches the relay node after the first signal and the second signal are mixed and subjected to noise addition; a target node receives a first mixed signal, wherein the first mixed signal is the signal that reaches the target node after the first signal and the second signal are mixed and subjected to noise addition; the relay node broadcasts the relay mixed signal; the target node receives a second mixed signal, wherein the second mixed signal is the signal that reaches the target node after the relay mixed signal is subjected to noise addition; and the target node acquires estimated values of the first signal and the second signal according to the first mixed signal and the second mixed signal. The encoding operation method effectively utilizes the interference produced after the signal mixing, improves the network capacity, and increases the bandwidth utilization rate.
Description
Technical field
The present invention relates to network coding technique, particularly relate to a kind of physical layer network code processing method, belong to communication technical field.
Background technology
Because the node in wireless network has mobility and portability, make the establishment of Wi-Fi conveniently, flexibly, wireless network has obtained general application.The link of wireless network has broadcast characteristic, and namely the node in wireless network can be to other a plurality of node broadcasts information, and same, the node in wireless network also can receive the information from other a plurality of nodes.Broadcast characteristic due to node, if a plurality of nodes in network are interior broadcast message at one time, can mutually mix in the free space of propagating between a plurality of signals that a plurality of node broadcasts are gone out, the mixing of signal can cause to produce between signal and disturb, and the information that causes receiving node to receive is inaccurate maybe can't receive required information.Therefore, be the generation that avoids interference, in prior art, the transmission method of information generally all adopts transmitted signal in succession to realize the transmission of information.
Fig. 1 is the structural representation of communication in the prior art wireless network; Fig. 2 is the sequential chart that in the prior art wireless network, in information transferring method, signal transmits.Have two source node S 1 and S2 in the assumed wireless network, via node R and destination node D, source node S 1 and S2 can be by via node R with communication to destination node D.As shown in Figure 2, in prior art, the communication process of source node S 1 and S2 and destination node D is as follows: at the first time slot T
10, source node S 1 broadcast message, the information of via node R reception sources node S1 broadcasting; At the second time slot T
20, via node R goes out the information broadcasting that the source node S 1 that observes is broadcasted, and is received the information of via node R broadcasting by destination node D, and destination node D obtains the information of source node S 1; At the 3rd time slot T
30, source node S 2 broadcast messages, the information of via node R reception sources node S2 broadcasting; At the 4th time slot T
40, via node R goes out the information broadcasting that the source node S 2 that observes is broadcasted, and is received the information of via node R broadcasting by destination node D, and destination node D obtains the information of source node S 2.Can find out, in prior art, two source nodes are during to same destination node transmission information, completing once that communication needs 4 time slots at least, communicate by letter if having between plural source node and destination node, and the time slot that needs will be more.
Can find out, in information transferring method, disturb for avoiding signal in the prior art wireless network, improve the accuracy of communication, when communicating by letter between a plurality of source nodes and same destination node, the time slot of completing primary information transmission needs is too much, make the network whole volume little, bandwidth availability ratio is low.
Summary of the invention
The purpose of this invention is to provide a kind of physical layer network code processing method, can effectively utilize signal to mix the rear interference that produces, improve bandwidth availability ratio, promote network capacity.
For achieving the above object, the invention provides a kind of physical layer network code processing method, comprising:
The first source node broadcasting first signal, the second source node broadcasting secondary signal;
Via node receives the relaying mixed signal, and described relaying mixed signal is that the signal that arrives described via node place after making an uproar is mixed, added to described first signal and secondary signal process; Destination node receives the first mixed signal, and described the first mixed signal is that the signal that arrives described destination node place after making an uproar is mixed, added to described first signal and secondary signal process;
The described relaying mixed signal of described via node broadcasting;
Described destination node receives the second mixed signal, and described the second mixed signal is described relaying mixed signal through adding the signal that arrives described destination node place after making an uproar;
Described destination node obtains the estimated value of first signal and secondary signal according to described the first mixed signal and the second mixed signal.
Wherein, described destination node comprises according to the estimated value of described the first mixed signal and the second mixed signal acquisition first signal and secondary signal:
With described the first mixed signal and the second mixed signal in conjunction with and obtain the observation signal Y of described destination node;
By Independent Component Analysis, described observation signal Y is processed, obtain the estimated value of described first signal and secondary signal.
Described described observation signal Y the processing by Independent Component Analysis comprises:
Described observation signal Y is carried out albefaction process, obtain whitened signal Z;
Described whitened signal Z is carried out analyzing and processing, obtain separation matrix V;
Obtain to comprise according to described separation matrix V and whitened signal Z the estimated matrix E:E=V that the estimated value by described first signal and secondary signal forms
HZ。
Describedly described observation signal Y carried out albefaction process, obtain whitened signal Z and comprise:
Obtain covariance matrix R according to described observation signal Y
Y, and according to described covariance matrix R
YObtain the albefaction matrix W;
By described albefaction matrix W, described observation signal Y is carried out albefaction and process, obtain whitened signal Z:Z=WY.
Described according to described observation signal Y acquisition covariance matrix R
Y, and according to described covariance matrix R
YObtaining the albefaction matrix W comprises:
Obtain the covariance matrix R of described observation signal Y according to described observation signal Y
Y, described covariance matrix R
YFor:
Wherein, Y=KX+N, X are the source signal matrix that first signal and secondary signal consist of, and K is the hybrid matrix of signal amplitude and phase fading coefficient on channel, and N is noise matrix; Y
HThe associate matrix of expression observation signal Y, R
XThe covariance matrix of expression source signal matrix X, R
NThe covariance matrix of expression noise matrix N,
With
Covariance matrix between expression source signal and noise, and
With
Be 0; σ
2Be the variance of noise matrix N, I
nBe n * n rank unit matrix;
To determinant | λ I-XX
H|=0 finds the solution, and obtains described covariance matrix R
YCharacteristic value { λ
1, λ
2..., λ
nAnd corresponding eigenvectors matrix A:A={a
1, a
2..., a
n, described characteristic value is that descending order is arranged;
Set matrix D: D=diag[(λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2)], wherein, diag () is that diagonal element is by (λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2) matrix that forms, { (λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2) be the characteristic value of observation signal Y;
Obtain described albefaction matrix W according to described matrix D and matrix A, described albefaction matrix W is: W=D
-1/2A
H=[(λ
1+ σ
2)
-1/2a
1, (λ
2+ σ
2)
-1/2a
2..., (λ
n+ σ
2)
-1/2a
n]
H
Described described whitened signal Z is carried out analyzing and processing, obtains separation matrix V and comprise:
According to described whitened signal Z, obtain the fourth order cumulant Q of described whitened signal Z
Z, and according to described fourth order cumulant Q
ZObtain described fourth order cumulant Q
ZFeature to { λ
r, M
r| 1≤r≤n}, according to described fourth order cumulant Q
ZFeature to obtaining characteristic set N
set: N
set={ λ
rM
r| 1≤r≤n}, wherein, λ
rBe characteristic value, M
rFor with λ
rThe matrix that corresponding characteristic vector forms, n is the number of observation signal;
By uniting diagonalization to described characteristic set N
setProcess, obtain described separation matrix V.
Described according to described fourth order cumulant Q
ZObtain described fourth order cumulant Q
ZFeature to { λ
r, M
r| 1≤r≤n} comprises:
With described fourth order cumulant Q
ZForm n
2* n
2Matrix Q;
Obtain characteristic value and the corresponding characteristic vector of described matrix Q according to described matrix Q, front n larger characteristic value and corresponding characteristic vector are formed described feature to { λ
r, M
r| 1≤r≤n}.
The described associating diagonalization that passes through is to described characteristic set N
setProcess, obtain described separation matrix V and comprise:
According to described characteristic set N
setTarget setting function C (V, N), described target function C (V, N) is:
Wherein, set
Diag () is by V for diagonal element
HN
rThe matrix that the characteristic value of V forms, a '
r, b '
r, c '
r, d '
rV
HN
rThe coefficient of V,
Be characteristic set N
setIn an element, i, l, k ∈ [1, n] is the subscript of observation signal;
Described target function C (V, N) is carried out iteration optimization process, obtain described separation matrix V.
Describedly described target function C (V, N) carried out iteration optimization process, obtain described separation matrix V and comprise:
Set matrix T according to target function C (V, N), matrix T is:
Wherein, a '
r-d '
r=(a
r-d
r) cos2 α+(b
r-c
r) sin2 α cos β+j (c
r-b
r) sin2 α sin β=P
Tg
r, P=[cos2 α, sin2 α cos β, sin2 α sin β]
TMatrix Re (G
HG) characteristic value characteristic of correspondence vector, g
r=[a
r-d
r, b
r+ c
r, j (c
r-b
r)]
T, r=1 ..., n, G=[g
1, g
2..., g
n]
T, Re () represents real;
Obtain Re (G
HG) the eigenvalue of maximum characteristic of correspondence of matrix vector, and according to described eigenvalue of maximum characteristic of correspondence vector sum formula P=[cos2 α, sin2 α cos β, sin2 α sin β]
TObtain corresponding α and β;
If Δ V
H, Δ V has the structure of hermitian, and
Carry out according to described α and β the updating value V that interative computation obtains separation matrix
new: V
new=V Δ V;
Updating value V according to described separation matrix
newUpgrade described eigenmatrix set N
setIn element N
r: N
r=Δ V
HN
rΔ V;
Judge whether α satisfies the iteration stopping condition, if the updating value of the separation matrix of this moment is described separation matrix V.
Further, described according to also comprising before target function C (V, N) setting matrix T:
The described separation matrix V=I of initialization
n
Set described iteration stopping condition, described iteration stopping condition is | sin α | and>1/100/ (n)
1/2
In technical solution of the present invention, the mixed signal transmission is adopted in signal transmission between source node and destination node, the time slot that makes the signal transmission take reduces, therefore, saved the time of network node busy channel, improve the throughput of network node, effectively improved the integrated communication capacity of network, promoted network bandwidth utilization factor; Simultaneously, by adopting the ICA isolation technics that the mixed signal that destination node receives is processed, make the separating resulting of mixed signal accurately, reliably in technical solution of the present invention.
Description of drawings
Fig. 1 is the structural representation of communication in the prior art wireless network;
Fig. 2 is the sequential chart that in the prior art wireless network, in information transferring method, signal transmits;
Fig. 3 is the structural representation of signal transmission in physical layer network code processing method of the present invention;
Fig. 4 is the sequential chart of signal transmission in physical layer network code processing method of the present invention;
Fig. 5 is the schematic flow sheet of physical layer network code processing method of the present invention;
Fig. 6 is that in physical layer network code processing method of the present invention, destination node is carried out the schematic flow sheet of separating treatment to mixed signal;
Fig. 7 is the schematic flow sheet that the present invention processes observation signal.
Embodiment
Below by drawings and Examples, technical scheme of the present invention is described in further detail.
Fig. 3 is the structural representation of signal transmission in physical layer network code processing method of the present invention; Fig. 4 is the sequential chart of signal transmission in physical layer network code processing method of the present invention.Embodiment of the present invention technical scheme can be applicable in existing wireless network, different from method for transmitting signals in prior art is, in the embodiment of the present invention, signal is to transmit with hybrid mode, the signal that destination node receives is that a plurality of source node signals are through mixed signal, destination node can be carried out separating treatment to the mixed signal that receives, and obtains the signal that source node sends.Particularly, Fig. 5 is the schematic flow sheet of physical layer network code processing method of the present invention.The method comprises:
Step 1, the first source node broadcasting first signal, the second source node broadcasting secondary signal;
Step 2, via node receive the relaying mixed signal, and described relaying mixed signal is that the signal that arrives described via node place after making an uproar is mixed, added to described first signal and secondary signal process; Destination node receives the first mixed signal, and described the first mixed signal is that the signal that arrives described destination node place after making an uproar is mixed, added to described first signal and secondary signal process;
Step 3, the described relaying mixed signal of described via node broadcasting;
Step 4, described destination node receive the second mixed signal, and described the second mixed signal is described relaying mixed signal through adding the signal that arrives described destination node place after making an uproar;
Step 5, described destination node obtain the estimated value of first signal and secondary signal according to described the first mixed signal and the second mixed signal.
In the present embodiment, as shown in Figure 4, at the first time slot T
1, the first source node S
1With the second source node S
2The signal that needs send is broadcasted away the first source node S
1First signal and second source node S of broadcasting
2The secondary signal of broadcasting is mixed in free space, is added and make an uproar, and first signal and secondary signal arrive respectively via node R and destination node D through the mixed signal of mixing, adding after making an uproar; Simultaneously, via node R and destination node D receive respectively relaying mixed signal and the first mixed signal; At the second time slot T
2, via node R is transmitted to destination node D by the relaying mixed signal that will receive with the form of broadcasting; Simultaneously, destination node D receives process that via node forwards add the second mixed signal after making an uproar in free space transmission.So far, the first source node S
1With the second source node S
2And a signal between destination node D is transmitted, destination node D carries out separating treatment to the first mixed signal and the second mixed signal that receive, can obtain the estimated value of first signal and secondary signal, and with this first signal and secondary signal estimated value respectively as the first source node S
1With the second source node S
2Send to the information of destination node.In the present embodiment technical scheme, source node can be also more than three or three, and correspondingly, the signal that purpose receives is that a plurality of source node signals mix, add the mixed signal after making an uproar, destination node can be separated it equally, obtains the estimated value of the broadcast singal of each source node.
Particularly, in the present embodiment, destination node can adopt Blind Signal Separation (Blind SignalSeparation is called for short BSS) interface differential technique to receive the first mixed signal and the second mixed signal are carried out separating treatment.The BSS technology be one under the known condition of observation signal, isolate the technology of source signal, it can be in the situation that any channel condition of the unknown, realizes mixing, adding the observation signal of making an uproar and separate through wireless channel what receiving node was received, obtains the primary signal that sending node sends.Therefore, destination node (receiving node) obtains by the mixed signal that receives is carried out separating treatment the primary signal that source node sends.Can find out, can realize mixed signal is separated by the BSS technology, make a plurality of source nodes can transmit a signal to simultaneously destination node, be that signal can carry out the transmission of information with the form of mixed signal, make each source node and destination node carry out the timeslot number of signal transmission few, therefore, time and channel that the signal transmission takies reduce accordingly, improve the speed of signal transmission, effectively improved the whole message capacity of network, promoted the utilance of the network bandwidth.
Fig. 6 is that in physical layer network code processing method of the present invention, destination node is carried out the schematic flow sheet of separating treatment to mixed signal.On the basis of above-mentioned Fig. 5 technical scheme, step 5 can specifically comprise:
In above-mentioned steps 51, destination node can be processed the first mixed signal and the second mixed signal that receive, and as the observation signal Y of destination node, observation signal Y also can regard the mixed signal of all signals that each source node sends as.
In above-mentioned steps 52, destination node is carried out analyzing and processing by independent component analysis (Independent ComponentAnalysis is called for short ICA) method to observation signal Y, and obtains the signal that source node sends.
Particularly, suppose that the source signal matrix of the signal composition that each source node sends is X, the source signal matrix in the present embodiment
X wherein
1[m] is the first source node S
1The first signal of broadcasting, x
2[m] is the second source node S
2The secondary signal of broadcasting, at this, also can be with x
1[m], x
2The first source node S that [m] obtains according to the ICA method as destination node
1With the second source node S
2Corresponding estimated value.Because wireless signal can be expressed as a discrete time function, therefore, the signal x[m that in the present embodiment, source node is broadcasted away] can be expressed as: x[m]=A
s[m] e
I θ s[m], A wherein
sBe the range value of signal, θ
sBe the phase place of signal, m is a number of samples that sends as the source node of signal source; The signal x[m that source node is broadcasted away] will be through decline, multipath and the phase shift of channel in spatial, so the signal that destination node (receiving node) receives can be expressed as x ' [m]: x ' [m]=k
ijA
s[m] e
I (θ s[m]+φ), k wherein
ij(i, j ∈ { s
1, s
2, r, d}) be signal from node j to the node i process amplitude attenuation factor on channel, and node j is to the distance dependent between node i, φ is the phase shift that signal produces on channel.
At the first time slot T
1, the first source node S
1With the second source node S
2The signal x of broadcasting
1[m], x
2Signal when [m] arrives via node R can be expressed as respectively:
Because signal mixes in the space, adds and make an uproar, the relaying mixed signal that finally receives of via node is:
Wherein, y
r[m] is the relaying mixed signal, n
r[m] is the noisy signal in source node arrival via node process.
Similarly, the first mixed signal of receiving of destination node is:
Wherein,
Be the first mixed signal,
It is the first source node S
1The first signal x of broadcasting
1Expression when [m] arrives destination node D,
It is the second source node S
2The secondary signal x of broadcasting
2Expression when [m] arrives destination node D, n
1[m] is the noise signal in source node signal arrival destination node process.
At the second time slot T
2, the relaying mixed signal y of via node R broadcasting
r[m] arrives the signal at destination node D place for passing through the signal that adds after making an uproar, and the second mixed signal of receiving of destination node D is:
Wherein,
Be the second mixed signal, n
2[m]=n
r[m]+n '
2[m] to be the source node signal arrive noisy signal in the destination node process, n ' by via node
2[m] is the noisy signal in relaying mixed signal arrival destination node process.
Therefore, the observation signal Y that receives of destination node is:
And Y=KX+N is arranged, wherein,
Be the hybrid matrix of signal amplitude and phase fading coefficient on channel,
Add the matrix of making an uproar for signal on channel.
Can find out, only require that solving hybrid matrix K just can utilize above-mentioned Formula For Solving to go out X, can obtain the first source node S
1The first signal value x of broadcasting
1[m] and the second source node S
2The secondary signal value x of broadcasting
2[m].
Particularly, Fig. 7 is the schematic flow sheet that the present invention processes observation signal.On the basis of above-mentioned Fig. 6 technical scheme, in order to solve hybrid matrix K, the present embodiment carries out albefaction and separating treatment with observation signal Y, utilizes albefaction matrix W and separation matrix V to come estimated mixing matrix K, thereby obtains the first source node S
1With the second source node S
2The source signal matrix X that sends.Step 52 can specifically comprise:
In above-mentioned steps 520, observation signal Y is carried out albefaction process, obtain whitened signal Z and comprise: obtain covariance matrix R according to described observation signal Y
Y, and according to described covariance matrix R
YObtain the albefaction matrix W; By described albefaction matrix W, described observation signal Y is carried out albefaction and process, obtain whitened signal Z:Z=WY.Wherein, obtain covariance matrix R according to observation signal Y
Y, and according to covariance matrix R
YObtaining the albefaction matrix W can specifically comprise:
Steps A 1, obtain the covariance matrix R of described observation signal Y according to described observation signal Y
Y, described covariance matrix R
YFor:
Wherein, Y=KX+N, X are the source signal matrix that first signal and secondary signal consist of, and K is the hybrid matrix of signal amplitude and phase fading coefficient on channel, and N is noise matrix; Y
HThe associate matrix of expression observation signal Y, R
XThe covariance matrix of expression source signal matrix X, R
NThe covariance matrix of expression noise matrix N,
With
Covariance matrix between expression source signal and noise, and
With
Be 0; σ
2Be the variance of noise matrix N, I
nBe n * n rank unit matrix;
Steps A 2, to determinant | λ I-XX
H|=0 finds the solution, and obtains described covariance matrix R
YCharacteristic value { λ
1, λ
2..., λ
nAnd corresponding eigenvectors matrix A:A={a
1, a
2..., a
n, described characteristic value is that descending order is arranged;
Steps A 3, setting matrix D: D=diag ((λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2)), wherein, diag () is that diagonal element is by (λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2) matrix that forms, { (λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2) be the characteristic value of observation signal Y;
Steps A 4, obtain described albefaction matrix W according to described matrix D and matrix A, described albefaction matrix W is: W=D
-1/2A
H=[(λ
1+ σ
2)
-1/2a
1, (λ
2+ σ
2)
-1/2a
2..., (λ
n+ σ
2)
-1/2a
n]
H
In above-mentioned steps A1, the average of usually supposing noise matrix N is 0, and with the signal of source node broadcasting be incoherent, therefore, the covariance matrix between signal and noise equals 0, namely
Similarly,
And σ
2Variance for additivity white Gaussian (AWGN) noise matrix N.In steps A 2, to determinant | λ I-XX
H|=0 characteristic value that obtains after finding the solution is arranged according to order from big to small, n larger characteristic value { λ before getting in characteristic value
1, λ
2..., λ
nAnd the matrix A that forms of corresponding characteristic vector={ a
1, a
2..., a
n, simultaneously, the characteristic value that can obtain observation signal matrix Y according to the character of matrix is { (λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2).
In above-mentioned steps 521, whitened signal Z is carried out analyzing and processing, obtaining separation matrix V can specifically comprise:
Step B1, according to described whitened signal Z, obtain the fourth order cumulant Q of described whitened signal Z
Z, and according to described fourth order cumulant Q
ZObtain described fourth order cumulant Q
ZFeature to { λ
r, M
r| 1≤r≤n}, according to described fourth order cumulant Q
ZFeature to obtaining characteristic set N
set: N
set={ λ
rM
r| 1≤r≤n}, wherein, λ
rBe characteristic value, M
rFor with λ
rThe matrix that corresponding characteristic vector forms, n is the number of observation signal, in the present embodiment, n is 2;
Step B2, by the associating diagonalization to described characteristic set N
setProcess, obtain described separation matrix V.
Wherein, in step B1, according to described fourth order cumulant Q
ZObtain fourth order cumulant Q
ZFeature to { λ
r, M
r| 1≤r≤n} comprises: with fourth order cumulant Q
ZForm n
2* n
2Matrix Q; Obtain characteristic value and the corresponding characteristic vector of matrix Q according to matrix Q, with front n larger characteristic value and corresponding characteristic vector composition characteristic to { λ
r, M
r| 1≤r≤n}.Particularly, if a random vector has zero-mean m
0With the covariance matrix R of unit (can be also that the unit covariance matrix is multiplied by a normal value variances sigma again
2), claim that this random vector is albefaction, therefore the vector after albefaction is processed satisfies following formula:
Wherein I is unit matrix.Generally, the covariance of establishing source signal matrix X is 1, i.e. E{XX
H}=1, synchronous signal X ' is the signal after processing through albefaction, namely
I=WR
X′W=WKE[XX
H]K
HW
H
=WKK
HW
H
Obviously, there is unitary matrice U=WK, satisfies unitary matrice definition I=UU
HSo, hybrid matrix K can be expressed as
K=W
-U=W
-[u
1,...,u
n]
Matrix W herein
-Represent the generalized inverse matrix of albefaction matrix W.
Processing through albefaction the signal Z that obtains is:
Z=WY=W(KX+N)=VX+WN
It is linear that signal Z after albefaction remains.According to formula E=V
HZ can obtain the expression formula of estimated matrix E:
E=V
HZ=V
HUX+V
HWN
If U=V knows V by the unitary matrice definition
HU=V
HV=I, I herein has identical dimension with V.The signal that is mixed with the AWGN noise that estimated matrix E represents is equal to the signal of source signal through producing after certain phase shift, so has high-order intersection cumulant identically vanishing to set up.
Therefore, in the present embodiment, the method for fourth order cumulant is adopted in blind separation for the whitened signal Z that produces after processing through albefaction.The four-dimensional cumulant Q of definable whitened signal Z
ZFor:
By to described fourth order cumulant Q
zCarry out computing, described fourth order cumulant Q
zFor:
Wherein, z
j *Be z
jConjugation, z
l *Be z
lConjugation, i, l, k ∈ [1, n] is the subscript of observation signal.
As can be known, Q
ZIn contain n
4Individual value is according to this fourth order cumulant
, with described fourth order cumulant Q
ZThe n that comprises
4Individual value is mapped to n
2* n
2Matrix Q, described matrix Q is:
Q finds the solution to matrix, obtains characteristic value and the corresponding characteristic vector of matrix Q, and with the characteristic value that obtains by from big to small order sequence, get wherein before n larger characteristic value, with n
2Individual characteristic vector consists of the matrix on a n * n rank, obtains Q
ZFeature to { λ
r, M
r| 1≤r≤n}, thus eigenmatrix set N obtained
set:
N
set={λ
rM
r|1≤r≤n}
Wherein, N
setThe set that is formed by n matrix.
In above-mentioned steps B2, by uniting diagonalization to characteristic set N
setProcess, obtain separation matrix V and comprise: according to described characteristic set N
setTarget setting function C (V, N), described target function C (V, N) is:
Wherein, set
Diag () is by V for diagonal element
HN
rThe matrix that the characteristic value of V forms, a '
r, b '
r, c '
r, d '
rV
HN
rThe coefficient of V,
Be characteristic set N
setIn an element, i, l, k ∈ [1, n] is the subscript of observation signal; Target function C (V, N) is carried out iteration optimization process, obtain separation matrix V.Wherein, target function C (V, N) is carried out iteration optimization process, obtain separation matrix V and comprise: set matrix T according to target function C (V, N), matrix T is:
Wherein, a '
r-d '
r=(a
r-d
r) cos2 α+(b
r-c
r) sin2 α cos β+j (c
r-b
r) sin2 α sin β=P
Tg
r, P=[cos2 α, sin2 α cos β, sin2 α sin β]
TMatrix Re (G
HG) characteristic value characteristic of correspondence vector, g
r=[a
r-d
r, b
r+ c
r, j (c
r-b
r)]
T, r=1 ..., n, G=[g
1, g
2..., g
n]
T, Re () represents real; Obtain Re (G
HG) the eigenvalue of maximum characteristic of correspondence of matrix vector, and according to this eigenvalue of maximum characteristic of correspondence vector sum formula P=[cos2 α, sin2 α cos β, sin2 α sin β]
TObtain corresponding α and β; If Δ V
H, Δ V has the structure of hermitian, and
Carry out according to described α and β the updating value V that interative computation obtains separation matrix
new: V
new=V Δ V; Updating value V according to described separation matrix
newUpgrade described eigenmatrix set N
setIn element N
r: N
r=Δ V
HN
rΔ V; Judge whether α satisfies the iteration stopping condition, if the updating value of the separation matrix of this moment is described separation matrix V.In addition, described according to also comprising before target function C (V, N) setting matrix T: the described separation matrix V=I of initialization
nSet described iteration stopping condition, described iteration stopping condition is | sin α | and>1/100/ (n)
1/2
Particularly, according to the characteristic set N of above-mentioned acquisition
set, with N
setIn an element representation be N
r:
And target setting function C (V, N):
And V
HN
rV has expression-form:
Wherein, diag () is by V for diagonal element
HN
rThe matrix that the characteristic value of V forms, a '
r, b '
r, c '
r, d '
rV
HN
rThe coefficient of V.
The iteration optimization process of target function C (V, N) just is equivalent to finds out α and β makes ∑
r| a '
r|
2+ | d '
r|
2Obtain maximum.According to 2 (| a '
r|
2+ | d '
r|
2)=| a '
r-d '
r|
2+ | a '
r+ d '
r|
2And N '
r=V
HN
rThe characteristic that after the V unitary transformation, matrix trace remains unchanged as can be known, to ∑
r| a
r|
2+ | d
r|
2Maximizing just is equivalent to ∑
r| a '
r|
2+ | d '
r|
2Maximizing, and then be equivalent to maximizing:
By above-mentioned formula N '
r=V
HN
rV, by calculating as can be known:
a′
r-d '
r=(a
r-d
r) cos2 α+(b
r-c
r) sin2 α cos β+j (c
r-b
r) sin2 α sin β is for ease of understanding and explanation, setting is following auxiliary vectorial:
P=[cos2α,sin2αcosβ,sin2αsinβ]
T
g
r=[a
r-d
r,b
r+c
r,j(c
r-b
r)]
T
G=[g
1,g
2,…,g
n]
T
Formula a '
r-d '
r=(a
r-d
r) cos2 α+(b
r-c
r) sin2 α cos β+j (c
r-b
r) sin2 α sin β can be expressed as:
a′
r-d′
r=P
Tg
r
So have:
If Δ V
HΔ V has the structure of hermitian (Hermitian), and Δ V can be made as:
Infinitely to approach with real source signal in order making through the estimated value of the source signal of the observation signal resulting separation that mixes, need to enter iterative process, iterative process each time is as long as calculate Re (G
HG) get final product.Wherein, iterative process is as follows:
At first carry out iteration initialization, at the iterative process initial phase, initialization separation matrix V is: V=I
n, and set the iteration stopping condition, described iteration stopping condition is | sin α |>1/100/ (n)
1/2Then enter iterative process, in iterative process each time, all need the characteristic value of compute matrix T, and get wherein maximum characteristic value and corresponding characteristic vector matrix G is upgraded, in iterative process, can be constantly to eigenmatrix set N
setIn each element Nr upgrade, its renewal process is as follows:
V
new=V·ΔV
N
r=ΔV
HN
rΔV
In the present embodiment, for to eigenmatrix set N
setCarry out diagonalization, will get maximum and as criterion, diagonal entry quadratic sum in objective function matrix be carried out iterative computation, and in the situation that the angle [alpha] that twice adjacent calculation obtains is more or less the same, stop iterative process, and then obtain separation matrix V.
In the present embodiment, after obtaining separation matrix V, utilize the V=U=WK that concerns between matrix W and matrix K, can obtain the expression formula of matrix K: K=W
-U=W
-V。
At last, utilize relational expression E=V
HZ can get:
At this moment, can obtain source signal x
1[m], x
2The estimated value e that [m] is corresponding
1[m], e
2[m], and with this estimated value as the source signal value, simultaneously can be to e
1[m], e
2[m] decodes, and obtains source signal x
1[m], x
2The information such as [m] and source node sign, and according to the information such as source node sign are sorted to source signal.
Need to prove, owing to can observation signal (mixed signal that receives) being separated the estimated value that obtains source signal by the physical layer encodes processing method, can obtain the source node sign after simultaneously estimated value being decoded, therefore can sort to source signal, overcome the source signal that obtains in the ICA method and had the fuzzy technological deficiency of order.In technique scheme of the present invention, source node can be for more than three or three, when the separation method of its method for transmitting signals and mixed signal and source node are two roughly the same.
Technical solution of the present invention is by adopting mixed signal to carry out the transmission of signal, the time slot that makes the signal transmission take reduces, therefore, saved the time of network node busy channel, improved the throughput of network node, effectively improve the integrated communication capacity of network, promoted network bandwidth utilization factor; Simultaneously, in technical solution of the present invention, by adopting the ICA isolation technics that mixed signal is processed, the source signal of acquisition accurately, reliably.
It should be noted that at last: above embodiment is only in order to technical scheme of the present invention to be described but not be limited, although with reference to preferred embodiment, the present invention is had been described in detail, those of ordinary skill in the art is to be understood that: it still can be modified or be equal to replacement technical scheme of the present invention, and these modifications or be equal to replacement and also can not make amended technical scheme break away from the spirit and scope of technical solution of the present invention.
Claims (9)
1. a physical layer network code processing method, is characterized in that, comprising:
The first source node broadcasting first signal, the second source node broadcasting secondary signal;
Via node receives the relaying mixed signal, and described relaying mixed signal is that the signal that arrives described via node place after making an uproar is mixed, added to described first signal and secondary signal process; Destination node receives the first mixed signal, and described the first mixed signal is that the signal that arrives described destination node place after making an uproar is mixed, added to described first signal and secondary signal process;
The described relaying mixed signal of described via node broadcasting;
Described destination node receives the second mixed signal, and described the second mixed signal is described relaying mixed signal through adding the signal that arrives described destination node place after making an uproar;
Described destination node obtains the estimated value of first signal and secondary signal according to described the first mixed signal and the second mixed signal, specifically comprise:
With described the first mixed signal and the second mixed signal in conjunction with and obtain the observation signal Y of described destination node;
By Independent Component Analysis, described observation signal Y is processed, obtain the estimated value of described first signal and secondary signal.
2. physical layer network code processing method according to claim 1, is characterized in that, described described observation signal Y the processing by Independent Component Analysis comprises:
Described observation signal Y is carried out albefaction process, obtain whitened signal Z;
Described whitened signal Z is carried out analyzing and processing, obtain separation matrix V;
Obtain to comprise according to described separation matrix V and whitened signal Z the estimated matrix E:E=V that the estimated value by described first signal and secondary signal forms
HZ。
3. physical layer network code processing method according to claim 2, is characterized in that, describedly described observation signal Y is carried out albefaction processes, and obtains whitened signal Z and comprise:
Obtain covariance matrix R according to described observation signal Y
Y, and according to described covariance matrix R
YObtain the albefaction matrix W;
By described albefaction matrix W, described observation signal Y is carried out albefaction and process, obtain whitened signal Z:Z=WY.
4. physical layer network code processing method according to claim 3, is characterized in that, and is described according to described observation signal Y acquisition covariance matrix R
Y, and according to described covariance matrix R
YObtaining the albefaction matrix W comprises:
Obtain the covariance matrix R of described observation signal Y according to described observation signal Y
Y, described covariance matrix R
YFor:
Wherein, Y=KX+N, X are the source signal matrix that first signal and secondary signal consist of, and K is the hybrid matrix of signal amplitude and phase fading coefficient on channel, and N is noise matrix; Y
HThe associate matrix of expression observation signal Y, R
XThe covariance matrix of expression source signal matrix X, R
NThe covariance matrix of expression noise matrix N,
With
Covariance matrix between expression source signal and noise, and
With
Be 0; σ
2Be the variance of noise matrix N, I
nBe n * n rank unit matrix, n is the number of observation signal;
To determinant | λ I-XX
H|=0 finds the solution, and obtains described covariance matrix R
YCharacteristic value { λ
1, λ
2..., λ
nAnd corresponding eigenvectors matrix A:A={a
1, a
2..., a
n, described characteristic value is that descending order is arranged;
Set matrix D: D=diag[(λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2)], wherein, diag () is that diagonal element is by (λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2) matrix that forms, { (λ
1+ σ
2), (λ
2+ σ
2) ..., (λ
n+ σ
2) be the characteristic value of observation signal Y;
Obtain described albefaction matrix W according to described matrix D and matrix A, described albefaction matrix W is:
W=D
-1/2A
H=[(λ
1+σ
2)
-1/2a
1,(λ
2+σ
2)
-1/2a
2,…,(λ
n+σ
2)
-1/2a
n]
H。
5. physical layer network code processing method according to claim 2, is characterized in that, described described whitened signal Z carried out analyzing and processing, obtains separation matrix V and comprise:
According to described whitened signal Z, obtain the fourth order cumulant Q of described whitened signal Z
Z, and according to described fourth order cumulant Q
ZObtain described fourth order cumulant Q
ZFeature to { λ
r, M
r| 1≤r≤n}, according to described fourth order cumulant Q
ZFeature to obtaining characteristic set N
set: N
set={ λ
rM
r| 1≤r≤n}, wherein, λ
rBe characteristic value, M
rFor with λ
rThe matrix that corresponding characteristic vector forms, n is the number of observation signal;
By uniting diagonalization to described characteristic set N
setProcess, obtain described separation matrix V.
6. physical layer network code processing method according to claim 5, is characterized in that, and is described according to described fourth order cumulant Q
ZObtain described fourth order cumulant Q
ZFeature to { λ
r, M
r| 1≤r≤n} comprises:
With described fourth order cumulant Q
ZForm n
2* n
2Matrix Q;
Obtain characteristic value and the corresponding characteristic vector of described matrix Q according to described matrix Q, front n larger characteristic value and corresponding characteristic vector are formed described feature to { λ
r, M
r| 1≤r≤n}.
7. physical layer network code processing method according to claim 5, is characterized in that, the described associating diagonalization that passes through is to described characteristic set N
setProcess, obtain described separation matrix V and comprise:
According to described characteristic set N
setTarget setting function C (V, N), described target function C (V, N) is:
Wherein, set
Diag () is by V for diagonal element
HN
rThe matrix that the characteristic value of V forms, a '
r, b '
r, c '
r, d '
rV
HN
rThe coefficient of V,
Be characteristic set N
setIn an element, i, l, k ∈ [1, n] is the subscript of observation signal; z
iBe i observation signal, z
lBe l observation signal, z
kBe k observation signal;
Described target function C (V, N) is carried out iteration optimization process, obtain described separation matrix V.
8. physical layer network code processing method according to claim 7, is characterized in that, describedly described target function C (V, N) is carried out iteration optimization processes, and obtains described separation matrix V and comprise:
Set matrix T according to target function C (V, N), matrix T is:
Wherein,
P=[cos2 α, sin2 α cos β, sin2 α sin β]
TMatrix Re (G
HG) characteristic value characteristic of correspondence vector, g
r=[a
r-d
r, b
r+ c
r, j (c
r-b
r)]
T, r=1 ..., n, G=[g
1, g
2..., g
n]
T, Re () represents real;
Obtain Re (G
HG) the eigenvalue of maximum characteristic of correspondence of matrix vector, and according to described eigenvalue of maximum characteristic of correspondence vector sum formula P=[cos2 α, sin2 α cos β, sin2 α sin β]
TObtain corresponding α and β;
If Δ V
H, Δ V has the structure of hermitian, and
Carry out according to described α and β the updating value V that interative computation obtains separation matrix
new: V
new=V Δ V;
Updating value V according to described separation matrix
newUpgrade described eigenmatrix set N
setIn element N
r: N
r=Δ V
HN
rΔ V;
Judge whether α satisfies the iteration stopping condition, if the updating value of the separation matrix of this moment is described separation matrix V.
9. physical layer network code processing method according to claim 8, is characterized in that, and is described according to also comprising before target function C (V, N) setting matrix T:
The described separation matrix V=I of initialization
n
Set described iteration stopping condition, described iteration stopping condition is | sin α | and>1/100/ (n)
1/2
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200910077201 CN101478374B (en) | 2009-01-19 | 2009-01-19 | Physical layer network code processing method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200910077201 CN101478374B (en) | 2009-01-19 | 2009-01-19 | Physical layer network code processing method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101478374A CN101478374A (en) | 2009-07-08 |
CN101478374B true CN101478374B (en) | 2013-06-12 |
Family
ID=40839010
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 200910077201 Expired - Fee Related CN101478374B (en) | 2009-01-19 | 2009-01-19 | Physical layer network code processing method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101478374B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101924605B (en) * | 2010-08-17 | 2012-12-05 | 重庆大学 | Double-hop cooperative transporting method based on physical-layer network coding |
CN102387000A (en) * | 2011-10-20 | 2012-03-21 | 中国空间技术研究院 | Network coding method based on blind signal separation |
CN105099618A (en) * | 2015-06-03 | 2015-11-25 | 香港中文大学深圳研究院 | Decoding method based on physical network coding and corresponding data processing method |
CN105207701B (en) * | 2015-08-04 | 2018-06-26 | 杭州电子科技大学 | For the coding/decoding method based on ICA in multiple cell multi-user multi-aerial system |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5706402A (en) * | 1994-11-29 | 1998-01-06 | The Salk Institute For Biological Studies | Blind signal processing system employing information maximization to recover unknown signals through unsupervised minimization of output redundancy |
CN101281584A (en) * | 2008-04-24 | 2008-10-08 | 湖南大学 | Method for recognizing radio frequency label based on ICA technique |
-
2009
- 2009-01-19 CN CN 200910077201 patent/CN101478374B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5706402A (en) * | 1994-11-29 | 1998-01-06 | The Salk Institute For Biological Studies | Blind signal processing system employing information maximization to recover unknown signals through unsupervised minimization of output redundancy |
CN101281584A (en) * | 2008-04-24 | 2008-10-08 | 湖南大学 | Method for recognizing radio frequency label based on ICA technique |
Non-Patent Citations (1)
Title |
---|
王福祥等.基于独立分量分析的高光谱图像端元提取算法.《仪器仪表学报》.2006,2077-2080. * |
Also Published As
Publication number | Publication date |
---|---|
CN101478374A (en) | 2009-07-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10020850B2 (en) | Master reference for base station network interface sourced from distributed antenna system | |
EP2075947B1 (en) | Virtual MIMO system method and base station | |
CN104735744B (en) | A kind of design method of the multi-hop relay routing based on terminal direct connection communication | |
CN101478374B (en) | Physical layer network code processing method | |
WO2004075459A3 (en) | Transmit diversity and separating multiple loopback signals | |
CN103039013B (en) | Utilize the method and apparatus of relay node operation intercommunication system | |
CN110289898A (en) | A kind of channel feedback method based on the perception of 1 bit compression in extensive mimo system | |
CN106028434A (en) | Dynamic interference guiding method | |
CN102510324B (en) | Signal transmission method based on network coding in multi-input and multi-output Y channel | |
CN105940625A (en) | Base station device, wireless communication system, and communication method | |
Alabed | Performance analysis of differential beamforming in decentralized networks | |
CN104333406A (en) | Decoding method of multi-cell, multi-user and multi-antenna system | |
CN103152086A (en) | Method for processing beam forming robustness signal on relay end of bi-direction multi-relay system | |
WO2021011147A3 (en) | Assisting communications of small data payloads with relay nodes | |
CN105706375A (en) | Topological pilot pollution eliminating method and device used for large-scale MIMO system | |
CN103701560A (en) | Precoding method based on interference alignment | |
CN101197639B (en) | Signal detection method and system, transmission device and receiving device | |
Alabed | Computationally efficient multi-antenna techniques for multi-user two-way wireless relay networks | |
CN108347271B (en) | Distributed interference alignment neutralization realization method capable of reaching maximum degree of freedom of system | |
CN111064501B (en) | Resource optimization method based on unmanned aerial vehicle double-relay communication system | |
CN105940700A (en) | Base station device, wireless communication system, and communication method | |
CN101753170A (en) | Multi-cell channel estimation method and device | |
CN103873125B (en) | The launching technique and equipment of downstream signal in a kind of multiaerial system | |
CN103220089B (en) | Based on the distributed cyclic overlap space-time code transmission method of two-way many junction networks | |
CN105049139A (en) | Cognitive User information transmission method applied to cognitive radio system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20130612 Termination date: 20200119 |
|
CF01 | Termination of patent right due to non-payment of annual fee |