CN105680903A - Periodic long-short code direct sequence spread spectrum code division multiple access signal multi-pseudo-code estimation method - Google Patents

Abstract

The invention relates to a periodic long-short code direct sequence spread spectrum code division multiple access signal multi-pseudo-code estimation method. The existing direct sequence spread spectrum code division multiple access signal pseudo code blind estimation technology cannot be applied to periodic long-short code direct sequence spread spectrum code division multiple access signals adopting short code spread spectrum and long code scrambling. The method comprises the steps of: constructing periodic long-short code direct sequence spread spectrum code division multiple access signals with complicated structures into a deficit matrix model of short code direct sequence spread spectrum code division multiple access signals, and modeling compound code matrix estimation as a blind source signal separation problem; applying a matrix filling theory to the compound code matrix estimation, and estimating user compound code sequences based on a singular value threshold algorithm and a fast independent component analysis algorithm; and finally proposing a delay triple correlation algorithm by utilizing a displacement stacking feature of an m sequence, and estimating long-short pseudo code sequences contained in the user compound code sequences. The periodic long-short code direct sequence spread spectrum code division multiple access signal multi-pseudo-code estimation method fully utilizes the matrix filling mathematic model and the triple correlation peak feature of the m sequence, and achieves blind estimation of the user compound code sequences, a long scrambling code sequence and a short spread spectrum code sequence of the signals.

Description

Cycle length code direct sequency-code division multiple access signal many pseudo-codes method of estimation

Technical field

The invention belongs to the blind parameter estirmation field of Direct Sequence Spread Spectrum Signal in communication antagonism, in particular to a kind of many pseudo-codes method of estimation using short code expansion frequency long code to add the cycle length code direct sequency-code division multiple access signal disturbed.

Background technology

Conventional communication system mainly utilizes the power of limited signal and bandwidth to transmit information as much as possible. Spread spectrum communication is the communication carried out data transmission in a bandwidth much larger than information speed, owing to it is not to save for the purpose of bandwidth, is thus different from conventional communication system.

Direct sequence spread spectrum (DirectSequenceSpreadSpectrum, DSSS) is one of main mode of spread spectrum technic. The core concept of direct-sequence communications system, before information code sequence sends, is modulated by signal by two-forty pseudo-random code, make the spread spectrum of signal, signal be submerged among noise so that direct sequence signal is difficult to detection. It is strong that direct sequence signal has interference rejection capability, the advantage such as be conducive to that signal is hidden, the access that can realize many locations, anti-fading ability are strong so that spread spectrum communication becomes the hot topic of research at present.

Code division multple access (CodeDivisionMultipleAccess, CDMA) is a kind of Digital Communications With Multiple Access mode, sets up channel by the code sequence of uniqueness. Direct sequency-code division multiple access (DS-CDMA) system realizes based on DSSS technology and CDMA technology. Direct sequency-code division multiple access signal can be divided into: short code direct sequency-code division multiple access signal, long code direct sequency-code division multiple access signal and use short code expand long code frequently and add the length code direct sequency-code division multiple access signal disturbed.

In communication antagonism, just due to advantages such as the immunity from interference of direct sequency-code division multiple access signal is strong, good concealment so that the detecting of direct sequency-code division multiple access signal and blind parameter estirmation when non-cooperative communication are quite difficult. In non-cooperative communication, pseudo-random code (abbreviation pseudo-code) estimation is prerequisite and the key of signal intercepting and capturing. The pseudo-code Estimation Study of short code direct sequency-code division multiple access signal is more ripe, and the research of long code direct sequency-code division multiple access signal has also obtained certain achievement.But the cycle, length code direct sequency-code division multiple access signal was due to its complex structure and multi-access inference, confidentiality is stronger, brings bigger difficulty and challenge to the pseudo-code blind estimate of non-cooperative communication.

Existing direct sequency-code division multiple access signal pseudo-code method of estimation realizes in conjunction with correlation matrix feature decomposition method, neural network method, matching matrix and three rank correlation methods mainly through blind separation. Owing to cycle length code direct sequency-code division multiple access signal comprising multiple user, each user comprises again two pseudo-codes, and correlation matrix feature decomposition method, neural network method and matching matrix need a large amount of signal sample, cycle length code direct sequency-code division multiple access signal all cannot be applicable to. Current many pseudo-codes blind estimate about cycle length code direct sequency-code division multiple access signal yet there are no disclosed correlative study achievement.

Summary of the invention

It is an object of the invention to for cannot the problem of each user's pseudo-code of blind estimate cycle length code direct sequency-code division multiple access signal in non-cooperative communication, propose a kind of cycle length code direct sequency-code division multiple access signal many pseudo-code blind estimating method relevant with three rank based on matrix fill-in, thus solve and be unable to estimate cycle length code direct sequency-code division multiple access signal each user many pseudo-codes problem.

Cycle length code direct sequency-code division multiple access signal many pseudo-codes blind estimating method in the present invention, comprises the following steps:

1, by cycle length code direct sequency-code division multiple access signal, to expand, code code sheet polydispersity index frequently is converted into baseband signal, the short code direct sequency-code division multiple access signal form of equal value of construction schedule length code direct sequency-code division multiple access signal, and sets up the disappearance matrix model of Received signal strength.

2, by signal deletion matrix being carried out matrix fill-in and singular value decomposition estimated signal compound numeral space.

3, signal compound numeral space carries out independent component analysis (Fast-ICA) estimate to obtain each user's compound code sequence.

4, calculating the triple correlation function of each user's compound code sequence respectively, be shifted superposition and three rank relevant peaks characteristics according to m-sequence, estimates successively to obtain each head of a household's scrambler sequence and short spreading code sequence.

The present invention makes full use of the constructional feature of cycle length code direct sequency-code division multiple access signal, builds short code direct sequency-code division multiple access signal model of equal value and blind source signal separation form, it is achieved the estimation of each user's compound code. Utilize displacement superposition and the three rank relevant peaks characteristics of long scrambler m-sequence in each user's compound code sequence, it is achieved to the estimation of the cycle length code direct sequency-code division multiple access signal many pseudo-codes of each user simultaneously.

Long scrambler and short spreading code are considered as compound code by the present invention, utilize the contact of cycle length code direct sequency-code division multiple access, long code direct sequency-code division multiple access and short code direct sequency-code division multiple access, construct short code direct sequency-code division multiple access signal form of equal value and Received signal strength disappearance matrix model, this disappearance matrix comprises the information of each user's compound code.

The present invention utilizes the singular value thresholding algorithm in matrix fill-in theory to lack matrix to received signal and fills, and carries out singular value decomposition to filling the full matrix obtained, can estimate to obtain compound numeral space.

The present invention estimates that the compound numeral space obtained not is the compound code matrix of cycle length code direct sequency-code division multiple access signal, it and signal compound code matrix belong to same subspace, there is linear transformation relation, utilize Fast-ICA algorithm to realize the blind separation of this linear model, thus estimate to obtain each user's compound code sequence.

The present invention estimates to comprise long scrambler sequence and short spreading code sequence in each user's compound code sequence obtained, compound code sequence loops moves to left some bit positions (expand frequently short code code sheet number is identical with the monocycle) be multiplied with former compound code sequence, it is possible to eliminate short spreading code to the impact of long scrambler triple correlation function.Utilize displacement superposition and the three rank relevant peaks characteristics of m-sequence, estimate successively to obtain long scrambler sequence and short spreading code sequence.

The invention has the beneficial effects as follows:

The disappearance matrix model that 1, cycle length code direct sequency-code division multiple access signal is configured to multi-user's short code direct sequency-code division multiple access signal, is modeled as blind source signal separation problem by signal compound code Matrix Estimation, can be simplified by sophisticated signal model by this modeling process.

2, matrix fill-in theory is applied to signal compound code Matrix Estimation, it is proposed to singular value thresholding algorithm and Fast-ICA algorithm can realize blind separation and the estimation of each user's compound code sequence.

3, eliminating short spreading code in compound code sequence by delay multiplication, for the impact of long scrambler, utilizes the displacement superposition of long code m-sequence to estimate the initial phase place of long code, it is possible to estimate long scrambler sequence and short spreading code sequence successively, and estimated performance improves greatly.

Embodiment

Further describe the implementation step of the present invention below.

Step 1, cycle length code direct sequency-code division multiple access signal many pseudo-codes method of estimation of the present invention in use, first by the cycle length code direct sequency-code division multiple access signal received to expand frequently after code code sheet polydispersity index, is converted into baseband signal y (l):

$y\left(l\right)=\underset{k=1}{\overset{K}{\Σ}}{A}_k{d}_k\left(l\right)b_k\left(l\right)c_k\left(l\right)+v\left(l\right)---\left(1\right)$

Wherein, l is sampling instant, l=0,1 ..., L-1; K is user's number; A_k、d_k(l)、b_k(l) and c_kL () represents that kth user's signal amplitude, information code sequence, cycle are that frequently code and cycle are the long scrambler of N for the expansion of G respectively, b_k(l) and c_kL () all adopts m-sequence and N > > G; V (l) is adding property white Gaussian noise; L is signal length, and assumes that L=JN, J are positive integer; Each is with containing per familyIndividual message digit,Represent and get the minimum integer being not less than x.

Signal length pseudo-code is considered as compound code, it is assumed that exist one containing same compound code sequence and information code sequence, expand K user's short code direct sequency-code division multiple access signal that frequency gain is N

$\begin{array}{c}\tilde{y}\left(l\right)=\underset{k=1}{\overset{K}{\Σ}}{A}_k\left(\underset{m=0}{\overset{M-1}{\Σ}}{d}_k\right(l\left)q\right(l-mN\left)\underset{j=0}{\overset{J-1}{\Σ}}{s}_k\right(l-jN\left)\right)+\tilde{v}\left(l\right)\\ =\tilde{x}\left(l\right)+\tilde{v}\left(l\right),l=0,1,\mathrm{...},\tilde{L}-1\end{array}---\left(2\right)$

Wherein,s_k(n)=b_k(n)c_k(n) (n=0,1 ..., N-1) and represent the compound code sequence of kth user,

The matrix form for N × M dimension can be represented:

$\tilde{Y}=SAD+\tilde{V}---\left(3\right)$

Wherein, S=[s₁s₂…s_K], s_k=[s_k(0)s_k(1)…s_k(N-1)]^T, D=[d₁d₂…d_K], d_k=[d_k(0)d_k(1)…d_k], (M-1) A=diag (A₁,A₂,…,A_K),For N × M ties up white Gaussian noise matrix, S is compound code matrix.

According toWithCorresponding relation, can obtain matrixIt is respectively with the element value of Y:

Wherein, (l)_NRepresent that l is to mould N complementation computing,Represent the maximum integer being not more than x.

Obviously, Received signal strengthIt isIn part sample, rest part be considered as disappearance. Therefore can be the mathematical model of disappearance matrix by K user's cycle length code direct sequency-code division multiple access signal modeling:

Wherein, Ω is the matrix of N × M dimension, andAll the other position elements are 0 (expression missing point); ⊙ represents matrix point multiplication operation; Y isEach Lieque has lost the disappearance matrix of about N-G specific position element.

By formula (5) and (6) by Received signal strengthIt is configured to the disappearance matrix Y of N × M dimension.

Step 2, from the matrix U that the left singularity characteristics vector that K the singular value that disappearance matrix Y estimation full matrix is maximum is corresponding forms_s, utilize singular value threshold value (SingularValueThresholding, the SVT) algorithm in matrix fill-in theory to realize, concrete grammar is:

1. initialize Z⁰∈R^N×M;

2. computational solution matrix X^j=F_τ(Z^j-1);

3. by Z^j=Z^j-1+δ(Y-X^j⊙ Ω) upgrade multiplier matrix Z^j;

If 4.2. then j=j+1, return; Otherwise, obtain estimated valueRightCarry out singular value decomposition and can obtain compound code sequence subspace U_sEstimated value

Wherein, Z⁰And Z^jFor being respectively Iterative Matrix initial value and intermediate value, X^jFor separating the iteration updated value of matrix.| | Y | |_FRepresent the Frobenius norm of matrix Y respectively.

F_τ(X) it is singular value threshold operator, it is defined as:

$F_{\mathrm{\τ}}\left(X\right)=\underset{n=1}{\overset{r}{\Σ}}{\mathrm{\λ}}_n{U}_z\left(n\right)V_z^H\left(n\right)---\left(7\right)$

Wherein, r is singular value number; λ_n(n=1,2, L, N) is for matrix Z is by the singular value of descending sort, and meets λ_r≤τ≤λ_r-1; U_z(n) and V_zN () represents the n-th row of the right singularity characteristics vector of left singularity characteristics vector sum of matrix Z respectively.

Step 3, U_sThe subspace opened with the column vector of compound code matrix S, the subspace opened of column vector belong to same subspace, namely there is the relation of linear transformation between them, utilize Fast-ICA algorithm, by U_sThe signal that direct blind separation goes out is S_f, to S_fCarry out symbolic operation can estimate to obtain compound code matrix, that is:

$\hat{S}=sign\left(S_f\right)---\left(8\right)$

The compound code matrix estimatedEach row be compound code sequence corresponding to each user

Step 4, by the compound code of kth userCirculation moves to right G bit position and is multiplied with former compound code, obtains

${\hat{a}}_k\left(n\right)={\hat{s}}_k\left(n\right){\hat{s}}_k(n+G)---\left(9\right)$

Calculate within the scope of N × NTriple correlation function

${\hat{C}}_{{\hat{a}}_k}(p,q)=\frac{1}{N}\underset{n=0}{\overset{N-1}{\Σ}}{\hat{a}}_k\left(n\right){\hat{a}}_k(n+p){\hat{a}}_k(n+q)---\left(10\right)$

Wherein, p and q is delay amount.

Due toThen according to m-sequence displacement superposition, can obtain:

${\hat{a}}_k\left(n\right)={\hat{c}}_k\left(n\right){\hat{c}}_k(n+G)={\hat{c}}_k(n+{\mathrm{\Γ}}_k)---\left(11\right)$

Wherein, Γ_kFor long code m-sequence delay amount.

According to m-sequence triple correlation function characteristicShould at coordinate point (G, Γ_k) place exists peak value. But due to noise and numerical evaluation error, Γ_kEstimator as follows:

${\mathrm{\Γ}}_k=\left\{q\right|{\hat{C}}_{{\hat{a}}_k}(p,q)=max\left\{{\hat{C}}_{{\hat{a}}_k}(p,q)\right\},p=G\}---\left(12\right)$

Circulation moves to right Γ_kBit position obtains long scrambler sequence

${\hat{c}}_k\left(n\right)={\hat{a}}_k(n-{\mathrm{\Γ}}_k),n=0,1,...,N-1---\left(13\right)$

According to the compound code sequence estimatedWith long scrambler sequenceCan estimate further to obtain short spreading code sequence

$\hat{b}\left(g\right)={\hat{s}}_k\left(g\right){\hat{c}}_k\left(g\right),g=0,1,...,G-1---\left(14\right)$

By step 4, according to estimating that the compound code sequence that obtains can estimate the respective long scrambler sequence of K user and short spreading code sequence respectively.

Above the preferred embodiments of the present invention and principle are described in detail, for the those of ordinary skill of this area, according to thought provided by the invention, embodiment will change, and these changes also should be considered as protection scope of the present invention.

Claims (6)

1. cycle length code direct sequency-code division multiple access signal many pseudo-codes method of estimation, it is characterised in that the method comprises the following steps:

(1) by cycle length code direct sequency-code division multiple access signal, to expand, code code sheet polydispersity index frequently is converted into baseband signal, the short code direct sequency-code division multiple access signal form of equal value of construction schedule length code direct sequency-code division multiple access signal, and set up the disappearance matrix model of Received signal strength;

(2) by signal deletion matrix being carried out matrix fill-in and singular value decomposition estimated signal compound numeral space;

(3) signal compound numeral space carries out independent component analysis (Fast-ICA) estimate to obtain each user's compound code sequence;

(4) calculating the triple correlation function of each user's compound code sequence respectively, be shifted superposition and three rank relevant peaks characteristics according to m-sequence, estimates successively to obtain each head of a household's scrambler sequence and short spreading code sequence.

2. cycle length code direct sequency-code division multiple access signal many pseudo-codes method of estimation as claimed in claim 1, it is characterized in that: the short code direct sequency-code division multiple access signal form of equal value of the construction schedule length code direct sequency-code division multiple access signal described in step (1) refers to and the length pseudo-code of Received signal strength y (l) is considered as special compound code, y (l) can represent and is:

$\begin{array}{c}y\left(l\right)=\underset{k=1}{\overset{K}{\Σ}}{A}_k\left(\underset{m=0}{\overset{M}{\Σ}}{d}_k\right(l\left)q\right(l-mG\left)\underset{j=0}{\overset{J}{\Σ}}{s}_k\right(l-jN\left)\right)+v\left(l\right)\\ =x\left(l\right)+v\left(l\right)\end{array}---\left(1\right)$

Wherein, l is sampling instant, l=0,1 ..., L-1; K is user's number; A_k、d_kL () represents kth user's signal amplitude, information code sequence respectively; s_k(n)=b_k(n)c_k(n) (n=0,1 ..., N-1) and represent the compound code sequence of kth user; X (l) represents useful signal, and v (l) is adding property white Gaussian noise;L is signal length, and assumes that L=JN, J are positive integer, 0≤j≤J;Each is with containing per familyIndividual message digit, N and G is respectively the cycle of long code and short code;

Assuming that existence one is containing K user's short code direct sequency-code division multiple access signal that same compound code sequence and information code sequence, expansion gain frequently are N

$\begin{array}{c}\tilde{y}\left(l\right)=\underset{k=1}{\overset{K}{\Σ}}{A}_k\left(\underset{m=0}{\overset{M-1}{\Σ}}{d}_k\right(l\left)q\right(l-mN\left)\underset{j=0}{\overset{J-1}{\Σ}}{s}_k\right(l-jN\left)\right)+\tilde{v}\left(l\right)\\ =\tilde{x}\left(l\right)+\tilde{v}\left(l\right),l=0,1,...,\tilde{L}-1\end{array}---\left(2\right)$

Wherein,

3. cycle length code direct sequency-code division multiple access signal many pseudo-codes method of estimation as claimed in claim 2, it is characterised in that: the disappearance matrix model of the structure Received signal strength described in step (1) refers to basisWithCorresponding relation:

Wherein (l)_NRepresent that l is to mould N complementation computing,Represent the maximum integer being not more than x;

Received signal strength is modeled as the mathematical model of disappearance matrix:

Wherein, Ω is the matrix of N × M dimension, andAll the other position elements are 0 (expression missing point); ⊙ represents matrix point multiplication operation; Y isEach Lieque has lost the disappearance matrix of about N-G specific position element.

4. cycle length code direct sequency-code division multiple access signal many pseudo-codes method of estimation as claimed in claim 3, it is characterized in that: described in step (2) signal deletion matrix is carried out matrix fill-in and singular value decomposition estimated signal compound numeral space refers to and utilizes singular value threshold value (SVT) algorithm fill cycle length code direct sequency-code division multiple access signal deletion matrix Y, obtain short code direct sequency-code division multiple access signal integrity matrix of equal valueEstimated valueRight againCarry out singular value decomposition, the matrix U of the left singularity characteristics vector composition that its K maximum singular value is corresponding_s∈R^N×KIt is signal compound numeral space.

5. cycle length code direct sequency-code division multiple access signal many pseudo-codes method of estimation as claimed in claim 4, it is characterised in that: step (3) utilizes independent component analysis (Fast-ICA) estimation to obtain each user's compound code sequence and refers to the signal compound numeral space U that step (2) obtains_s∈R^N×KThe subspace opened with the column vector of signal compound code matrix S, the subspace opened of column vector belong to same subspace, namely there is linear transformation T between them, that is:

U_s=TS (6)

For this linear mixed model can according to Fast-ICA algorithm, by U_sThe signal that direct blind separation goes out is S_f, to S_fCarry out symbolic operation can estimate to obtain compound code matrix, that is:

$\hat{S}=sign\left(S_f\right)---\left(7\right);$

The compound code matrix estimatedEach row be compound code sequence corresponding to each user

6. cycle length code direct sequency-code division multiple access signal many pseudo-codes method of estimation as claimed in claim 1, it is characterized in that: step (4) estimates length pseudo-code sequence successively according to compound code sequence, need to eliminating short spreading code impact by ring shift, recycling m-sequence displacement superimposed characteristics is estimated each with the initial phase place Γ of head of a household's scrambler sequence_k, Γ_kEstimator be:

${\mathrm{\Γ}}_k=\left\{q\right|{\hat{C}}_{{\hat{a}}_k}(p,q)=max\left\{{\hat{C}}_{{\hat{a}}_k}(p,q)\right\},p=G\}---\left(8\right)$

By the compound code of kth userCirculation moves to right G bit position and is multiplied with former compound code and to obtainCirculation moves to right Γ_kBit position obtains long scrambler sequence

c_k(n)=a_k(n-Γ_k), n=0,1 ..., N-1 (9)

According to the compound code sequence estimated and long scrambler sequence, it is possible to estimate further to obtain short spreading code sequence

${\hat{b}}_k\left(g\right)={\hat{s}}_k\left(g\right){\hat{c}}_k\left(g\right),g=0,1,...,G-1---\left(10\right).$