CN102271104B - Multi-user detection method and device of CDMA (Code Division Multiple Access) network, and multi-user detection filter - Google Patents

Abstract

The invention discloses a multi-user detection method and device of a CDMA (Code Division Multiple Access) network, and a multi-user detection filter, relating to the technical field of CDMA. The method comprises the following steps of: iteratively calculating a weighted vector of the multi-user detection filter based on a balanced constant modulus algorithm according to an input signal; and performing multi-user detection filter treatment on the input signal according to the weighted vector to obtain each user signal. Aiming at the disadvantages of the current SDCMA (Steepest Descent Constant Modulus Algorithm) and LSCMA (Least-Squares Constant Modulus Algorithm) algorithms, the multi-user detection method and the system of the CDMA network based on the improved constant modulus algorithm are provided, therefore, a target signal can be effectively converged under strong multi-access interference; the rapid convergence property of the LSCMA still can be kept under weak interference; the multi-user detection effect under the CDMA network can be effectively improved; and the user capacity of the CDMA network can be improved.

Description

Cdma network multi-user test method, device and Multiuser Detection filter

Technical field

The present invention relates to CDMA(Code Division Multiple Access, code division multiple access) technical field, relate in particular to a kind of cdma network multi-user test method, cdma network multi-user detection device and Multiuser Detection filter.

Background technology

Multiuser Detection is to suppress the effective way that multiple access disturbs, effectively promotes CDMA mobile communication system.Its basic thought is to utilize cross-correlation information between the user that multiple access comprises in disturbing to estimate, reduces or eliminate the influence of interference.Main multi-user test method has best Multiuser Detection, linear multi-user detection etc. at present.Though above non-blind Detecting has some advantages, when channel response changed suddenly or new cochannel interference occurs, training sequence need resend, and this will cause the waste of channel width.Therefore research does not need training, must not know that the blind adaptive multi-user detector of interference user information becomes the demand of practical application yet.

The blind adaptive multi-user detector can have accomplished in many ways, and wherein constant modulus algorithm (CMA:Constant Modulus Algorithm) is exactly one.The basic thought of CMA is to utilize the amplitude statistics characteristic that sends signal to carry out blind signal processing, keeps the constant coefficient of adjusting filter at receiving terminal by the amplitude that makes output signal.Consider the synchronous direct sequence spread spectrum cdma system of base band, suppose to have K user, its reception signal is:

$r\left(t\right)=\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{A}_k{b}_k{s}_k\left(t\right)+\mathrm{\δn}\left(t\right),t\∈[0,T]---\left(1\right)$

Wherein, A _k, b _k, s _k(t) represent k user's reception signal amplitude respectively, send symbol and normalization signature waveform.N (t) is additive white Gaussian noise, and variance is σ ²Here b _kBe the BPSK modulation signal, with equiprobability value ± 1.s _k(t) energy is 1 in [0, T],

$s_k\left(t\right)=\underset{i-1}{\overset{N-1}{\mathrm{\Σ}}}{c}_i^{k}p(t-{\mathrm{iT}}_c)---\left(2\right)$

Here N is spreading gain,

For the value of distributing to k user is ± 1 spreading code; P is the interior normalization chip of chip period (Tc) (chip) waveform, NTc=T.At receiving terminal, receive signal r (t) by behind the chip matched filter, with the spreading rate sampling, its output in a symbol period with vector representation is:

$r=\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{A}_k{b}_k{s}_k+\mathrm{\δn}=\mathrm{SAb}+\mathrm{\δn}---\left(3\right)$

Wherein, S=[s ₁..., s _K], s _kBe k user's normalized signal wave vector, A=diag(A ₁.A _K), b=[b ₁, L, b _k] ^TR=[r ₀..., r _N-1] ^TN=[n ₀..., n _N-1] ^TBe the white Gauss noise vector, and E{n}=0, E{nnT}=IN, δ ²It is the white Gaussian noise variance matrix.For detecting the signal of user k(1≤k≤K), then the weighted value of Multiuser Detection filter is w _k=[w _{K, 0}..., w _{K, N-1}] ^T, be output as y _k=w _k ^TR.

The realization the earliest of permanent mould adaptive algorithm is based on steepest descent algorithm (SDCMA:steepest descent Constant Modulus Algorithm), and the signal that hypothesis sends in constant modulus algorithm has constant envelope, and its cost function is defined as:

J(w(k))=E{||w ^H(k)r(k)| ^p-|α| ^p| ^q} （4）

α is the signal amplitude of array output end expectation in the formula.Utilize different p, q can produce multiple different steepest descent algorithm.Because the cost function of constant modulus algorithm is non-linear, so can't directly find the solution, can only adopt alternative manner to approach.When adopting direction of steepest descent to be optimized, its iterative formula is:

$w(n+1)=w\left(n\right)-{\mathrm{\μ}\▿}_{w}J\left(w\left(n\right)\right)---\left(5\right)$

Here μ〉0, be step factor,

The gradient operator of expression w replaces desired value with the up time value, makes p=1, q=2, and, the steepest decline constant modulus algorithm that obtains is called the 1-2 type.If make α=1, then the iterative equation of the steepest decline constant modulus algorithm of 1-2 type is:

w(n+1)=w(n)-μr _ne _n ^* （6）

Here

Be error function, y (n) is according to the n time iteration subscriber signal of calculating of w (n) as a result here.In addition, work as p=1, during q=1, Work as p=2, during q=1, e (n)=2y (n) sgn (| y (n) | ²-1); Work as p=2, during q=2, e (n)=4y (n) sgn (| y (n) | ²-1).The outstanding advantage of SDCMA algorithm is to realize simple.Because target function is close to quadratic function near minimal point, and steepest descent method to be linear approximation formula according to target function obtain, therefore this method convergence is slower.

Least square constant modulus algorithm (LSCMA:least-squares Constant Modulus Algorithm) basic principle is as follows, and the least square constant modulus algorithm minimizes permanent mould cost function:

J(w)=E[|w ^Hr|-1] ² （7）

Gauss method definition cost function according to expansion:

$F\left(w\right)=\underset{l=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{\θ}}_l^2={\left|\right|\mathrm{\θ}\left(w\right)\left|\right|}_2^2---\left(8\right)$

Then cost function has the quadratic sum form of part Taylor series expansion

Here d is bias vector, and U (w) is the Jacobian determinant of θ (w), and its form is

F (w+d) to the gradient of bias vector d is:

${\▿}_d\left(F(w+d)\right)=2\frac{\∂F(w+d)}{{\∂d}^{*}}=2\frac{\∂\left\{{[\mathrm{\θ}\left(w\right)+U^H\left(w\right)d]}^H\right[\mathrm{\θ}\left(w\right)+U^H\left(w\right)d\left]\right\}}{{\∂d}^{*}}$

$=2[U\left(w\right)\mathrm{\θ}\left(w\right)+u{\left(w\right)U}^H\left(w\right)d]---\left(9\right)$

Order

Can get d=-[U (w) U ^H(w)] ^-1U (w) θ (w), so w (n+1)=w (n)-[U (w (n)) U ^H(w (n))] ^-1U (w (n)) θ (w (n)), n represents the iteration step number here.Compare cost function J (w)=E[|w ^HR|-1] ²With

As can be known, θ (w)=| w ^TR|-1 utilizes above-mentioned derivation J (w)=E[|w as can be known ^HR|-1] ²Weight vector more new formula be:

w(n+1)=w(n)-(rr ^T) ^-1r[y(n)-φ(k)] ^*=(rr ^T) ^-1rφ ^*(n) (10)

Here $\mathrm{\φ}\left(n\right)={[\frac{y\left(1\right)}{\left|y\left(1\right)\right|},\frac{y\left(2\right)}{\left|y\left(2\right)\right|},\·\·\·,\frac{y\left(m\right)}{\left|y\left(m\right)\right|}]}^T.$

Because it is that the popularization of Gaussian processes designs constant modulus algorithm that the LSCMA algorithm has utilized non-linear least square, improved convergence of algorithm speed effectively, but in actual use, because Gaussian processes has certain requirement to the selection of initial point, if the minimal point of initial point and target function often causes the failure of algorithm, in addition at a distance of too far away, when receiving the taking on morbit forms property of correlation matrix of data, the LSCMA method can not well be used.

Summary of the invention

In view of above problem proposes the present invention.

The technical problem that the present invention will solve provides a kind of cdma network multi-user test method, has the advantage of the good and fast convergence rate of convergence.

According to an aspect of the present invention, provide a kind of cdma network multi-user test method, comprising: according to the weight vectors of input signal based on balanced constant modulus algorithm iterative computation Multiuser Detection filter; According to weight vectors input signal is carried out Multiuser Detection filtering and handle the signal that obtains each user; Wherein, balanced constant modulus algorithm adopts following formula that the weight vectors of Multiuser Detection filter is carried out iteration:

$w_i(n+1)=w_i\left(n\right)-(\frac{1}{1+\exp \left(\mathrm{\α}\right|e\left(n\right)\left|\right)}{R}_{\mathrm{rr}}{\left(n\right)}^{-1}+\frac{1}{1+\exp (-\mathrm{\α}|e\left(n\right)\left|\right)}\mathrm{\μI})r_{n}e\left(n\right),\mathrm{\α}>0,\mathrm{\μ}>0,$

Wherein, n represents iteration step number, w _i(n) be i user's weight vectors w _iValue after the n time iteration, μ is step factor, the error function when e (n) is the n time interative computation, r _nBe the reception vector signal that is used for the n time iteration use that wireless receiver receives, R _Rr(n) for receiving signal r _nCovariance matrix, I is unit matrix.In the process of the weight vectors that input signal is carried out iterative computation Multiuser Detection filter based on balanced constant modulus algorithm, if

|w _i(n+1)-w _i(n)|≤δ

Judge that then weight vectors restrains; Wherein δ is predetermined threshold value.

Preferably, this method also comprises: carry out the spreading rate sampling to received signal by the chip matched filter and obtain input signal.

Preferably, this method also comprises: the signal to each user of obtaining is adjudicated.

The technical problem that the present invention will solve provides a kind of Multiuser Detection filter, has the advantage of the good and fast convergence rate of convergence.

According to a further aspect in the invention, provide a kind of Multiuser Detection filter, comprising: balanced constant modulus algorithm module is used for according to the weight vectors of input signal based on balanced constant modulus algorithm iterative computation Multiuser Detection filter; The multiple user signals acquisition module is used for according to weight vectors input signal being carried out Multiuser Detection filtering and handles the signal that obtains each user; Wherein, balanced constant modulus algorithm module adopts following formula that the weight vectors of Multiuser Detection filter is carried out iteration:

$w_i(n+1)=w_i\left(n\right)-(\frac{1}{1+\exp \left(\mathrm{\α}\right|e\left(n\right)\left|\right)}{R}_{\mathrm{rr}}{\left(n\right)}^{-1}+\frac{1}{1+\exp (-\mathrm{\α}|e\left(n\right)\left|\right)}\mathrm{\μI})r_{n}e\left(n\right),\mathrm{\α}>0,\mathrm{\μ}>0,$

Wherein, n represents iteration step number, w _i(n) be i user's weight vectors w _iValue after the n time iteration, μ is predefined step factor, but will satisfy μ〉0 condition, the error function when e (n) is the n time interative computation, r _nBe the reception vector signal that is used for the n time iteration use that wireless receiver receives, R _Rr(n) for receiving signal r _nCovariance matrix, I is unit matrix.Balanced constant modulus algorithm module is when determining | w _i(n+1)-w _i(n) | the judgement weight vectors is restrained during≤δ; Wherein δ is predetermined threshold value.

The technical problem that the present invention will solve provides a kind of cdma network multi-user detection device, has the advantage of the good and fast convergence rate of convergence.

According to another aspect of the invention, provide a kind of cdma network multi-user detection device, comprise above-mentioned Multiuser Detection filter.

Preferably, this device also comprises: the chip matched filter is used for carrying out to received signal the spreading rate sampling and obtains input signal.

Preferably, this device also comprises: judge module is used for each user's of obtaining signal is adjudicated.

CDMA multi-user test method provided by the invention, device and Multiuser Detection filter, can under disturbing, strong multiple access effectively converge to echo signal, and under weak jamming, still can keep least square constant modulus algorithm characteristic of rapid convergence, have the advantage of the good and fast convergence rate of convergence.

Description of drawings

Fig. 1 illustrates the flow chart according to an embodiment of cdma network multi-user test method of the present invention;

Fig. 2 illustrates the flow chart according to another embodiment of cdma network multi-user test method of the present invention;

Fig. 3 illustrates the structure chart according to an embodiment of cdma network multi-user detection device of the present invention.

Embodiment

With reference to the accompanying drawings the present invention is described more fully, exemplary embodiment of the present invention wherein is described.

The inventor notices, LSCMA is that with the main distinction of the method that detects the account form of iteration weighted value w (n) is different with SDCMA more than two kinds, in order to take full advantage of the advantage of two kinds of algorithms, can the iteration weight of two kinds of algorithms be weighted, the iteration weighted value of two kinds of algorithms is controlled by error function e (n).Based on above-mentioned thought, special proposition multi-user test method of the present invention.

Fig. 1 illustrates the flow chart according to an embodiment of cdma network multi-user test method of the present invention.

As shown in Figure 1, in step 102, according to the weight vectors of input signal based on balanced constant modulus algorithm iterative computation Multiuser Detection filter.Can iteration threshold perhaps be set for iteration preestablishes iterations, perhaps in conjunction with both.

Balanced constant modulus algorithm adopts following formula that the weight vectors of Multiuser Detection filter is carried out iteration:

$w_i(n+1)=w_i\left(n\right)-(\frac{1}{1+\exp \left(\mathrm{\α}\right|e\left(n\right)\left|\right)}{R}_{\mathrm{rr}}{\left(n\right)}^{-1}+\frac{1}{1+\exp (-\mathrm{\α}|e\left(n\right)\left|\right)}\mathrm{\μI})r_{n}e\left(n\right),\mathrm{\α}>0,\mathrm{\μ}>0,$

Wherein, n represents iteration step number, w _i(n) be i user's weight vectors w _iValue after the n time iteration, μ is predefined step factor, but will satisfy μ〉0 condition, the error function when e (n) is the n time interative computation, r _nBe the reception vector signal that is used for the n time iteration use that wireless receiver receives, R _Rr(n) for receiving signal r _nCovariance matrix, I is unit matrix.

In step 104, according to weight vectors input signal is carried out Multiuser Detection filtering and handle the signal that obtains each user.

In above-described embodiment, bigger in the starting stage of iteration error, at this moment mainly adopt the iteration step length of SDCMA to carry out iteration, along with reducing of error function, the weight that receives the correlation matrix of data becomes big LSCMA algorithm gradually and becomes to take as the leading factor.Algorithm both can utilize the SDCMA algorithm to suppress stronger interference like this, can converge to echo signal rapidly again in the process of algorithmic statement.In addition, the method for the present invention's proposition and the complexity of LSCMA algorithm all are O (N ³), do not promote the complexity of practical application, so have actual using value.

Fig. 2 illustrates the flow chart according to another embodiment of cdma network multi-user test method of the present invention.

As shown in Figure 2, step 201 is carried out spreading rate to received signal by the chip matched filter and is adopted the acquisition input signal.Receive signal

Comprise K subscriber signal, receive signal r (t) by behind the chip matched filter, with the spreading rate sampling, its output signal in a symbol period with vector representation is

S=[s wherein ₁..., s _K], s _kBe k user's normalized signal wave vector, A=diag(A ₁.A _K), b=[b ₁..., b _k] ^TR=[r ₀..., r _N-1] ^TN=[n ₀..., n _N-1] ^TBe the white Gauss noise vector, and E{n}=0, E{nn ^T}=I _N, δ ²It is the white Gaussian noise variance matrix.The output signal of chip matched filter is as follow-up input signal.

Step 202 adopts balanced constant modulus algorithm input signal to be carried out the weight vectors of iterative computation Multiuser Detection filter.If adopting the weight vectors of the Multiuser Detection filter of balanced constant modulus algorithm is w=[w ₀..., w _N-1] ^T, (detection of 1≤i≤K), then the weight vectors of corresponding balanced constant modulus algorithm Multiuser Detection filter is w to existing hypothesis to user i _i=[w _{I, 0}..., w _{I, N-1}] ^T, adopt balanced constant modulus algorithm that input signal is carried out iteration and realize weighted value w _iCalculating, make w here _i(n) expression w _iValue after the n time iterative computation, w _iIteration adopt balanced constant modulus algorithm:

$w_i(n+1)=w_i\left(n\right)-(\frac{1}{1+\exp \left(\mathrm{\α}\right|e\left(n\right)\left|\right)}{R}_{\mathrm{rr}}{\left(n\right)}^{-1}+\frac{1}{1+\exp (-\mathrm{\α}|e\left(n\right)\left|\right)}\mathrm{\μI})r_{n}e\left(n\right),\mathrm{\α}>0,\mathrm{\μ}>0,$

Step 203, judge | w _i(n+1)-w _i(n) |≤δ? if in the iterative process | w _i(n+1)-w _i(n) |≤δ, then judge weighted vector value w _iRestrain, continue step 204, proceed interative computation otherwise return step 202.δ is predetermined threshold value, can be according to the actual conditions setting.

Step 204 is carried out Multiuser Detection filtering according to the weight vectors of Multiuser Detection filter to input signal and is handled the signal that obtains each user.According to satisfying of each user | w _i(n+1)-w _i(n) | the weight vectors w of the Multiuser Detection filter of≤δ _i(n) vector r carries out the filtering processing to received signal, obtains the signal y at user i _i=w _i ^TR.

Step 205 is adjudicated filtered each user's signal.For example, adopt existing Technology of Judgment to carry out the judgement of each user's signal.

Fig. 3 illustrates the structure chart according to an embodiment of cdma network multi-user detection device of the present invention.As shown in Figure 3, the cdma network multi-user detection device comprises chip matched filter 31, Multiuser Detection filter 32 and judge module 33 among this embodiment.Multiuser Detection filter 32 comprises balanced constant modulus algorithm module 321 and multiple user signals acquisition module 322.Chip matched filter 31 r (t) to received signal carries out spreading rate sampling acquisition input signal.Multiuser Detection filter 32 carries out Multiuser Detection filtering based on balanced constant modulus algorithm to input signal and handles the signal that obtains each user.Each user's of 33 pairs of acquisitions of judging module signal is adjudicated.Balanced constant modulus algorithm module 321 is according to the weight vectors of input signal based on balanced constant modulus algorithm iterative computation Multiuser Detection filter; Multiple user signals acquisition module 322 carries out Multiuser Detection filtering according to weight vectors to input signal and handles the signal that obtains each user; Wherein, balanced constant modulus algorithm module 321 adopts following formula that the weight vectors of Multiuser Detection filter is carried out iteration:

$w_i(n+1)=w_i\left(n\right)-(\frac{1}{1+\exp \left(\mathrm{\α}\right|e\left(n\right)\left|\right)}{R}_{\mathrm{rr}}{\left(n\right)}^{-1}+\frac{1}{1+\exp (-\mathrm{\α}|e\left(n\right)\left|\right)}\mathrm{\μI})r_{n}e\left(n\right),\mathrm{\α}>0,\mathrm{\μ}>0,$

Wherein, n represents iteration step number, w _i(n) be i user's weight vectors w _iValue after the n time iteration, μ is predefined step factor, but will satisfy μ〉0 condition, the error function when e (n) is the n time interative computation, r _nBe the reception vector signal that is used for the n time iteration use that wireless receiver receives, R _Rr(n) for receiving signal r _nCovariance matrix, I is unit matrix.

In one embodiment, balanced constant modulus algorithm module is when determining | w _i(n+1)-w _i(n) | the judgement weight vectors is restrained during≤δ; Wherein δ is predetermined threshold value.

The deficiency that the present invention is directed to existing SDCMA and LSCMA algorithm proposes to realize cdma network multi-user test method and system based on improving constant modulus algorithm, can under disturbing, strong multiple access effectively converge to echo signal, and under weak jamming, still can keep least square constant modulus algorithm characteristic of rapid convergence, can effectively improve the effect of Multiuser Detection under the cdma network, improve the user capacity of cdma network.

Of the present inventionly realize cdma network multi-user test method and system based on constant modulus algorithm, LSCMA algorithm and SDCMA algorithm combined, the advantage of two kinds of multi-user test methods can effectively be obtained and avoid separately deficiency therefore have very strong validity.

Description of the invention provides for example with for the purpose of describing, and is not exhaustively or limit the invention to disclosed form.Many modifications and variations are obvious for the ordinary skill in the art.Selecting and describing embodiment is for better explanation principle of the present invention and practical application, thereby and makes those of ordinary skill in the art can understand the various embodiment that have various modifications that the present invention's design is suitable for special-purpose.

Claims (7)

1. a cdma network multi-user test method is characterized in that, comprising:

According to the weight vectors of input signal based on balanced constant modulus algorithm iterative computation Multiuser Detection filter;

According to described weight vectors input signal is carried out Multiuser Detection filtering and handle the signal that obtains each user;

Wherein, described balanced constant modulus algorithm adopts following formula that the weight vectors of Multiuser Detection filter is carried out iteration:

Wherein, n represents iteration step number, w _i(n) be i user's weight vectors w _iValue after the n time iteration, μ is step factor, the error function when e (n) is the n time interative computation, r _nBe the reception vector signal that is used for the n time iteration use that wireless receiver receives, R _Rr(n) for receiving signal r _nCovariance matrix, I is unit matrix;

In the process of the weight vectors that input signal is carried out iterative computation Multiuser Detection filter based on balanced constant modulus algorithm, if

|w _i(n+1)-w _i(n)|≤δ

Judge that then weight vectors restrains; Wherein δ is predetermined threshold value.

2. method according to claim 1 is characterized in that, also comprises:

Carry out the spreading rate sampling to received signal by the chip matched filter and obtain described input signal.

3. method according to claim 1 is characterized in that, also comprises:

Signal to each user of obtaining is adjudicated.

4. a Multiuser Detection filter is characterized in that, comprising:

Balanced constant modulus algorithm module is used for according to the weight vectors of input signal based on balanced constant modulus algorithm iterative computation Multiuser Detection filter;

The multiple user signals acquisition module is used for according to described weight vectors input signal being carried out Multiuser Detection filtering and handles the signal that obtains each user;

Wherein, described balanced constant modulus algorithm module adopts following formula that the weight vectors of Multiuser Detection filter is carried out iteration:

Wherein, n represents iteration step number, w _i(n) be i user's weight vectors w _iValue after the n time iteration, μ is step factor, the error function when e (n) is the n time interative computation, r _nBe the reception vector signal that is used for the n time iteration use that wireless receiver receives, R _Rr(n) for receiving signal r _nCovariance matrix, I is unit matrix;

Described balanced constant modulus algorithm module is when determining

|w _i(n+1)-w _i(n)|≤δ

The time judge that weight vectors restrains; Wherein δ is predetermined threshold value.

5. a cdma network multi-user detection device is characterized in that, comprises Multiuser Detection filter as claimed in claim 4.

6. multi-user detection device according to claim 5 is characterized in that, also comprises:

The chip matched filter is used for carrying out to received signal the spreading rate sampling and obtains described input signal.

7. multi-user detection device according to claim 6 is characterized in that, also comprises:

Judge module is used for each user's of obtaining signal is adjudicated.

Images