CN101465668A - Method for detecting space-time unite signal - Google Patents

Abstract

The invention discloses a space-time combined signal detection method which includes that space interference cross-correlation matrix Rn is determined on a receiver end through channel estimation; a Rank-N approximate matrix E of the space interference cross-correlation matrix Rn is computed; an orthogonal projection matrix R of the space interference cross-correlation matrix Rn is determined according to the approximate matrix E; the orthogonal projection matrix R is used as the inverse matrix of the space interference cross-correlation matrix R for detecting sent signals; wherein, N is positive integer. The space-time combined signal detection method is adopted to conveniently realize the whitening of interference and noise; meanwhile, the target signal is ensured not to be whitened.

Description

A kind of method for detecting space-time unite signal

Technical field

The present invention relates to the signal detection technique in the wireless communication system, particularly a kind of method for detecting space-time unite signal.

Background technology

In the TD-SCDMA system of M transmit antennas, N root reception antenna, suppose to have K VRU sign indicating number, L=W-1, wherein, W is long for the channel window, and for the m root antenna of receiving terminal, its mixed channel matrix is,

$B^m={\left[\begin{array}{cccc}{b}_1^m& b_2^m& \·\·\·& b_K^m\end{array}\right]}_{(16+L)\×K}=\left[\begin{array}{c}{C^m}_{16\×K}\\ {{\tilde{D}}^m}_{L\×K}\end{array}\right]$

Wherein, b ^mBe the convolution of VRU sign indicating number and wireless channel, suppose W≤16,, promptly obtain zero padding to 32 row under the B partitioning of matrix

${\tilde{B}}^m={\left[\begin{array}{ccc}\begin{array}{cc}{b}_1^m& b_2^m\\ 0& 0\end{array}& \·\·\·& \begin{array}{c}{b}_K^m\\ 0\end{array}\end{array}\right]}_{32\×K}=\left[\begin{array}{c}{C^m}_{16\×K}\\ {D^m}_{16\×K}\end{array}\right]$

In above-mentioned TD-SCDMA system, the receiving symbol vector on the receiving terminal m root antenna can be described by following formula,

$x_{16(N+1)\×1}^m=A_{16(N+1)\×\mathrm{KN}}^m{d}_{\mathrm{KN}\×1}+n_{16(N+1)\×1}^m$

Wherein,

Be the sytem matrix on the m root reception antenna, d _{KN * 1}Be the transmission symbolic vector of N transmit antennas,

It is the noise vector of m root reception antenna.Here, the modulation symbol of t code word formation with all VRU sign indicating numbers then sends symbolic vector x _{KN * 1}In have N modulation symbol, comprise K code word in each symbols.

If consider then to have following system description by M root reception antenna simultaneously:

x＝Ad+n

x＝[x ^1T?x ^2T...x ^MT] ^T

A＝[A ^1T?A ^2T...A ^MT] ^T(2)

n＝[n ^1T?n ^2T...n ^MT] ^T

Wherein, x is the receiving symbol vector, and H is a sytem matrix, and d is for sending symbolic vector, and n is a noise matrix, for receiving symbol vector y as the formula (2), utilize the MMSE-BLE algorithm that it is carried out input at receiving terminal to be,

$\tilde{d}={(A^H{R}_n^{-1}A+R_d^{-1})}^{-1}{A}^H{R}_n^{-1}x---\left(3\right)$

Wherein, R _dThe autocorrelation matrix of representative information symbol generally is similar to unit matrix, R _nRepresent space-time interference matrix, generally can be decomposed into the relevant battle array of time domain R _tRelevant battle array R with the spatial domain _sKronecker long-pending, the time covariance matrix generally can be taken as unit matrix, i.e. R in the practical communication environment _t=I.Therefore can get, $R_n={\mathrm{\σ}}_t{I}_{\mathrm{KN}}\⊗R_s,$ Wherein,

Represent Kronecker long-pending.Here, N represents the number of modulation symbol.

Therefore, (3) formula finally can be written as

$\tilde{d}={(A^H{R}_s^{-1}A+{\mathrm{\σ}}_t{I}_{\mathrm{KN}})}^{-1}{A}^H{R}_s^{-1}X---\left(4\right)$

On the other hand, if adopt the ZF-BLE algorithm, its separate into

$\tilde{d}={\left(A^H{R}_s^{-1}A\right)}^{-1}{A}^H{R}_s^{-1}x---\left(5\right).$

So, can utilize above-mentioned formula (4) or (5) to estimate the transmission signal of transmitting terminal, finish input.

When utilizing formula (4) or (5) to carry out input, need at first to characteristics of radio channels h _uEstimate, and to spatial interference cross-correlation matrix R _sEstimate; Then according to aforesaid characteristics of radio channels h _uDetermine A with the relation of channel characteristics matrix A; Utilize formula (4) or (5) to estimate the transmission signal of transmitting terminal at last.

Wherein, to characteristics of radio channels h _uWhen estimating, try to achieve according to the Midamble signal that receives the user.Concrete estimation procedure comprises: suppose that K user is h=(h to the channel impulse response of a certain antenna of antenna array ^{(1) T}, h ^{(2) T}..., h ^{(K) T}) ^T, then the Midamble signal of every antenna reception can be expressed as: y=Gh+n _Mid, wherein: $G=\left(\begin{array}{c}{m}_1,m_{128},....,m_2\\ m_2,m_1,...,m_3\\ ......................\\ m_{128},.....m_2,m_1\end{array}\right)$ Be the convolution matrix that basic Midamble sign indicating number m constitutes, size is the 128*128 dimension; n _MidFor being superimposed upon the noise vector on this antenna.Thereby have:

$y(1:128)=m(1:128)\⊕h(1:128),$ The expression circular convolution.So:

h(1:128)＝IFFT(FFT(y)/FFT(m))。(6)

Channel estimation errors is bigger to the influence of ZF-BLE algorithm, for reducing channel estimation errors, will handle initial channel estimating.The channel impulse response that estimation on each antenna is obtained is handled according to the methods below:

P1＝0；

For?n＝1：128

if ${10}^{\mathrm{Threshold}\_\mathrm{PP}/10}*N_0>\frac{1}{M}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{\left(\mathrm{abs}\left(h_{\mathrm{av}}^m\left(n\right)\right)\right)}^2$

$h_i^m\left(n\right)=0;$

${h_n}^m\left(n\right)=h_{\mathrm{av}}^m\left(n\right);$

else

${h_i}^m\left(n\right)=h_{\mathrm{av}}^m\left(n\right);$

h _n ^m(n)＝0；

P1++；

end；

end； (7)

Wherein, M is the reception antenna number,

The original channel of the up subscriber signal that calculates for formula (6) estimates,

Be channel estimating through the up subscriber signal of reprocessing, Be the channel weights of interference/noise signal, N ₀Be this time slot chip noise power, can try to achieve that Threshold_PP is specified by base station control station according to idle window or spatial noise covariance.The physical significance of Threshold_PP can be learnt by simple derivation: ${10}^{\mathrm{Threshold}\_\mathrm{PP}/10}*N_0>\frac{1}{M}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{\left(\mathrm{abs}\left(h_{\mathrm{av}}^m\left(n\right)\right)\right)}^2$

Take the logarithm in both sides,

$10\log \frac{\frac{1}{M}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left(\mathrm{abs}\left(h_{\mathrm{av}}^m\left(n\right)\right)\right)}^2}{N_0}<\mathrm{Threshold}\_\mathrm{PP}.$

What Threshold_PP represented as can be known is the physical significance of signal to noise ratio, if promptly signal to noise ratio is greater than a certain threshold T hreshold_PP, then keeps the channel estimating of this point, otherwise is changed to 0.If the channel impulse response h that handles on pairing each antenna of certain user of back is 0, do not consider this user when then carrying out joint-detection; Can be got by above-mentioned derivation, P1 represents the number of all active paths.

Specifically to spatial interference cross-correlation matrix R _sThe process of estimating comprises: according to what form in the above-mentioned channel estimation process

Structural matrix T, $T=\left[\begin{array}{c}{\left({h_n}^1\right)}^T\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {\left({h_n}^M\right)}^T\end{array}\right],$ Disturb cross-correlation matrix to be according to the matrix T estimation space again: $R_s=\frac{\mathrm{PT}\&CenterDot;T^H}{(P-P1)D},---\left(8\right)$

Wherein, P is all the possible path numbers in the channel estimation process, is specially the code length of midamble sign indicating number, promptly 128; P1 is all active path numbers after the channel estimating, and just channel impulse response is not 0 number of path, and D is a constant, with the difference of compensation frequency domain energy and time domain energy.

When utilizing formula (4) or (5) to estimate the transmission signal of transmitting terminal, spatial interference cross-correlation matrix R _sThe spatial characteristics that reflects presence of intercell interference.And, according to this formula (4) or (5) as seen, to handle during at present for sky, the decorrelation of relevant interference is to disturb cross-correlation matrix R by usage space _sContrary the realization, just in formula (4) or (5), comprise This decorrelation is handled will disturb with noise and is changed the white signal in time domain and spatial domain into, guarantees that the result that space-time unite detects is that linear optimal is estimated.

But, the characteristic of spatial interference signal, i.e. spatial interference cross-correlation matrix changes with the variation of Complex Channel environment, and also the variation with the wireless system condition of work changes.On the one hand, under the environment of N frequency point networking, co-channel interference dominates.Especially under urban area circumstance, because the area coverage of sub-district (sector) is quite little, the spacing of co-frequency cell is also quite near; Add long and narrow street and can embody the effect of similar waveguide sometimes again, the transmission attenuation of wireless signal is diminished, transmission range is farther.Therefore, very how far closely different the antenna for base station of each sub-district can receive the subscriber signal of co-frequency cell.These different capacity signals are received by the antenna of Target cell by various very complicated transmission channels and different transmission ranges, and consequently the dimension of the spatial interference matrix of Target cell becomes quite big.On the other hand, when mobile phone switched in the minizone, the power of certain mobile phone might increase fast, usually can cause interference signal association correlation matrix dimension decline situation.The situation of above-mentioned two aspects all causes spatial interference cross-correlation matrix R _sIt is very difficult that the numerical operation of directly inverting becomes.Adopt fixed-point computation matrix R _sInvert and may occur because of the restriction of dynamic range calculating the situation of overflowing, make that this operation result appearance is wrong and become unavailable, thereby cause the demodulation signal interference ratio gain severe exacerbation of final joint-detection, the error rate be increased sharply.The deterioration of up reception signal interference ratio influences the mistake adjustment of uplink power control conversely, makes mobile phone terminal improve constantly up transmitting, and makes the situation of whole network become bad more, until sub-district collapse, mobile phone call drop.

In addition, even the technology that the base station adopts multi-plot joint to detect is considered as useful signal with a lot of adjacent areas co-channel interference signal and detects, thereby the interference of the actual inhibition in base station is greatly reduced.But, because the technology computational complexity that multi-plot joint detects is quite high, the storage resource demands height, the restriction of base station equipment cost, and the restriction of the equation dimension of least squares equation (4) or (5), total accessible code channel number that multi-plot joint detects is limited, and the code channel number of co-channel interference that can participate in joint-detection is also just limited.That is to say that co-channel interference can't be eliminated fully, its influence always exists.More in addition, if the co-channel interference signal can't be eliminated or suppress fully, so, the relative alien frequencies of the level of the interference signal of whole network system is higher, and up transmitting increases relatively.At this moment, the reprocessing thresholding Threshold_PP of channel estimating chooses and can't reach optimum, and precision of channel estimation is also limited, and the uplink interference signal that causes channel estimating to provide is infiltrated the echo signal of this sub-district, may cause echo signal is carried out albefaction, also will cause system's demodulation performance to descend.

Summary of the invention

In view of this, the invention provides the prewhitening method in a kind of space-time unite detection, can realize easily disturbing with noise and carry out albefaction, guarantee simultaneously echo signal not to be carried out albefaction.

For achieving the above object, the present invention adopts following to technical scheme:

A kind of method for detecting space-time unite signal comprises:

Determine spatial interference cross-correlation matrix R at receiving terminal by channel estimating _s

Calculate described spatial interference cross-correlation matrix R _sRank-N approximate matrix E, and determine spatial interference cross-correlation matrix R according to this approximate matrix E _sOrthogonal intersection cast shadow matrix

With this orthogonal intersection cast shadow matrix

As spatial interference cross-correlation matrix R _sInverse matrix, send the detection of signal, described N is a positive integer.

Preferably, described computer memory disturbs cross-correlation matrix R _nRank-N approximate matrix E comprise:

Utilize power method and progressively the null method computer memory disturb cross-correlation matrix R _sN eigenvalue of maximum, and calculate N corresponding respectively principal eigenvector e of this N characteristic value _i, wherein, the value of i is 1 to N;

Utilize described N principal eigenvector structure approximate matrix E=[e _i] _I=1:N

Preferably, described utilize power method and progressively the null method computer memory disturb cross-correlation matrix R _sN eigenvalue of maximum, and calculate N corresponding respectively principal eigenvector e of this N characteristic value _iComprise:

A, with spatial interference cross-correlation matrix R _sAs current matrix;

B, utilize characteristic value of current matrix computations and corresponding principal eigenvector thereof, and upgrade current matrix;

Whether the characteristic value number that c, judgement obtain is N, if, then continue to carry out the described operation that utilizes N principal eigenvector structure approximate matrix, otherwise, step b returned.

Preferably, describedly utilize characteristic value of current matrix computations and corresponding principal eigenvector thereof, and upgrade current matrix and comprise:

B1, make that current principal eigenvector is e _i=[1,1 ..., 1] ^T/ sqrt (M), and wherein [1,1 ..., 1] be M dimension unit vector, M is the reception antenna number;

B2, the current principal eigenvector of renewal are $e_i=\frac{R_i*e_i}{{\left|\right|R_i*e_i\left|\right|}_2},$ Wherein, ‖ ‖ ₂Be 2 rank norms;

B3, according to current matrix and the current principal eigenvector after upgrading to calculate current characteristic value be a _i=e _i ^T* R _i* e _i/ (e _i ^T* e _i);

B4, execution in step b2～b3 finishes iteration one time successively, up to reaching default iterations, carries out next step;

B5, the current characteristic value that obtains according to step b4 and current principal eigenvector upgrade current matrix and are $R_i=R_i-a_i*e_i*e_i^T.$

Preferably, describedly determine spatial interference cross-correlation matrix R according to this approximate matrix E _sOrthogonal intersection cast shadow matrix

Comprise:

Wherein, E ^TTransposed matrix for E.

Preferably, the signal to noise ratio according to computation complexity, system resource and demodulation requires to determine N.

Preferably, describedly determine spatial interference cross-correlation R by channel estimating at receiving terminal _sComprise:

From the subscriber signal that receives, extract the data of training sequence part, utilize this data estimation original user up-converter channel impact responding h _i

According to interference power Piip and reprocessing thresholding threshold_pp, carry out the noise reduction reprocessing, obtain interference/noise channel weight matrix T;

According to described interference/noise channel weight matrix T, calculate the spatial interference cross-correlation matrix R of up reception _s

Preferably, the spatial interference cross-correlation matrix R of the up reception of described calculating _sComprise:

$R_s=\frac{\mathrm{PT}\&CenterDot;T^H}{(P-P1)},$ Wherein, P is the code length of described training sequence, and P1 is the active path number that described noise reduction reprocessing obtains.

As seen from the above technical solution, among the present invention, determine spatial interference cross-correlation matrix R by channel estimating at receiving terminal _sThen, computer memory disturbs cross-correlation matrix R _sRank-N approximate matrix E, calculate the orthogonal intersection cast shadow matrix of this approximate matrix E at last And with this orthogonal intersection cast shadow matrix

As the inverse matrix of space-time interference matrix, send the detection of signal.By the way, utilize the orthogonal intersection cast shadow matrix of space-time interference matrix to carry out input, avoided matrix inversion operation, greatly reduce the computation complexity of input, be suitable for fixed-point calculation as the inverse matrix of space-time interference matrix; Simultaneously, utilize the approximate matrix of low-rank approximate matrix, simplified matrix dimension, and avoid that this sub-district echo signal is carried out albefaction and suppress as space-time interference matrix.

Description of drawings

Fig. 1 is a wireless communication system illustraton of model of the present invention.

Fig. 2 is the particular flow sheet of method for detecting space-time unite signal provided by the invention.

Fig. 3 is under the N frequency point networking, according to the present invention with the performance that obtains of the signal detecting method of background technology schematic diagram relatively.

Embodiment

For making purpose of the present invention, technological means and advantage clearer,, the present invention is described in further details below in conjunction with accompanying drawing.

In the present invention, carried out modeling again for wireless system.Concrete system model as shown in Figure 1, wherein, the uplink user data d of this sub-district is received by this cell base station through wireless channel H, simultaneously, the uplink user data I of co-frequency neighbor cell is received by this cell base station through wireless channel S.

In this system model, suppose that alien frequencies disturbs and thermal noise all is an additive white Gaussian, so, the signal that this cell base station receives can be written as: x=Ad+Si+e (9),

Wherein, e is that alien frequencies disturbs and the thermal noise sum.

Usually, we can suppose that wireless channel A and S are irrelevant, and their codomain is no crosslinked.In addition, can think that co-channel interference signal Si and alien frequencies disturb and thermal noise e is a quadrature, therefore can obtain:

$\&lang;\mathrm{Si},e\&rang;=0\&DoubleRightArrow;i^H{S}^{H}e=0,$ $\&ForAll;i\&NotEqual;0\&DoubleRightArrow;S^{H}e=0,$

In like manner, can think that also signal that this sub-district is subjected to and alien frequencies disturb and thermal noise e is a quadrature, i.e. A ^HE=0.

Suppose that channel matrix H and S are known,, can use the orthogonal intersection cast shadow matrix of spatial interference cross-correlation matrix in order to estimate this cell uplink user's data

Multiply by antenna for base station and receive formula (9), obtain:

(10); According to the notion of spatial decomposition as can be known, all spatial interference signals that comprise co-channel interference are formed spatial interference cross-correlation space, and the rectangular projection space of spatial interference cross-correlation matrix and spatial interference cross-correlation space are mutually orthogonal, also just with all spatial interference signal in orthogonal that comprise co-channel interference, so orthogonal intersection cast shadow matrix of spatial interference cross-correlation matrix

With the product of co-channel interference signal Si be 0, promptly

Simultaneously, as described in the background art, co-channel interference is main interference signal, therefore has

Further can get

Formula (11) and (12) substitution (10) are obtained

Formula (13) equal sign the right and left is taken advantage of A together ^H, and utilize A ^HE=0 just obtains the estimation of this cell uplink user data:

Further, can obtain the least mean-square error of this community user data is estimated:

According to formula (4) and (14) more as can be seen, adopting the contrary of spatial interference cross-correlation matrix and its orthogonal intersection cast shadow matrix is of equal value substantially.

On the other hand, utilize the knowledge of signal subspace, can utilize Low-Rank (low-rank) approximate matrix of spatial interference cross-correlation matrix to realize for the calculating of the orthogonal intersection cast shadow matrix of spatial interference cross-correlation matrix.

Based on above-mentioned analysis, basic thought of the present invention is: the inverse matrix of utilizing the orthogonal intersection cast shadow matrix of spatial interference cross-correlation matrix to be used as the spatial interference cross-correlation matrix is carried out input, and utilizes Low-Rank (low-rank) approximate matrix of spatial interference cross-correlation matrix to determine the orthogonal intersection cast shadow matrix of spatial interference cross-correlation matrix.

Fig. 2 is the particular flow sheet of method for detecting space-time unite signal provided by the invention.As shown in Figure 2, this method comprises:

Step 201, receiving terminal receives data, to characteristics of radio channels h _uCarry out channel estimating and reprocessing.

The process of describing in the operation of this step and the background technology is identical, just repeats no more here.

Step 202 utilizes the interference/noise channel weight matrix T computer memory of reprocessing to disturb cross-correlation matrix R _s

In this step, utilize formula (8) computer memory in the background technology to disturb cross-correlation matrix R _s, when utilizing formula (8) to calculate, also can not consider the influence of constant D, even $R_s=\frac{\mathrm{PT}\&CenterDot;T^H}{(P-P1)}.$

Step 203, utilize power method and progressively null method calculate spatial interference cross-correlation matrix R _sN maximum characteristic value, and calculate N characteristic value N principal eigenvector e of correspondence respectively _i

Usually, one time power method obtains calculating a characteristic value and corresponding principal eigenvector, in the method for the invention, calculates N principal eigenvector of N characteristic value and difference correspondence by the successive iteration of power method.

Step 204 utilizes N the principal eigenvector that obtains in the step 203 to form the Low-Rank approximate matrix E=[e of spatial interference cross-correlation matrix _i] _{I=1: N}

Step 205 is disturbed the orthogonal intersection cast shadow matrix of cross-correlation matrix according to the Low-Rank approximate matrix E computer memory of spatial interference cross-correlation matrix

Realized utilizing the Low-Rank approximate matrix of space cross-correlation matrix to determine the purpose of the orthogonal intersection cast shadow matrix of space cross-correlation matrix by above-mentioned steps 203～204.In the said process, N gets positive integer, and its value is determined according to computation complexity, system resource and demodulation effect.Concrete, the N value is big more, and the computation complexity of step 203～204 is then big relatively more, and the system resource of occupying is many more, and the demodulation effect of its realization is good more.Wherein, be that Rank-1 is approximate during the simplest N=1,

Step 206 is with orthogonal intersection cast shadow matrix

As spatial interference cross-correlation matrix R _nThe contrary input of carrying out.

In this step, utilize orthogonal intersection cast shadow matrix

Carry out input.Concrete, approximate calculation MMSE-BLE separate for

Perhaps, approximate calculation ZF-BLE separate for

Wherein, σ _tThe expression noise/interference power, I _KNBe the unit matrix of K*N dimension, K is a number of users, and N is the number of modulation symbol.

So far, method for detecting space-time unite signal flow process provided by the invention finishes.In above-mentioned flow process, the operation in the step 203 specifically can may further comprise the steps:

A, with the spatial interference cross-correlation matrix as current matrix;

B, utilize characteristic value of current matrix computations and corresponding principal eigenvector thereof, and upgrade current matrix;

In this step, concrete computation of characteristic values and principal eigenvector and the operation of upgrading current matrix comprise:

B1, make that current principal eigenvector is e _i=[1,1 ..., 1] ^T/ sqrt (M), and wherein [1,1 ..., 1] be M dimension unit vector, M is the reception antenna number, i is the index of current main characteristic vector;

B2, the current principal eigenvector of renewal are $e_i=\frac{R_i*e_i}{{\left|\right|R_i*e_i\left|\right|}_2},$ Wherein, ‖ ‖ ₂Be 2 rank norms;

B3, according to current matrix and the current principal eigenvector after upgrading to calculate current dominant eigenvalue be a _i=e _i ^T* R _i* e _i/ (e _i ^T* e _i);

B4, execution in step b2～b3 finishes iteration one time successively, up to reaching default iterations N _i, carry out next step;

Iterations in this step can determine that this iterations is big more according to current available resource, and the resource that takies is many more, but low-rank approximate matrix that obtains and spatial interference cross-correlation matrix are close more.Simultaneously, in the computational process for N characteristic value, when calculating the principal eigenvector of different dominant eigenvalue correspondences, its iterations is generally identical, can certainly be different.Preferably, this iterations can be set to 16.

B5, the current dominant eigenvalue that obtains according to step b4 and current principal eigenvector upgrade current matrix and are $R_i=R_i-a_i*e_i*e_i^T.$

So far, finish the calculating of a characteristic value and principal eigenvector, and current matrix is upgraded.

Whether the characteristic value number that c, judgement obtain is N, if, then continue to carry out the described operation that utilizes N principal eigenvector structure approximate matrix, otherwise, step b returned.

The operation of above-mentioned steps a～c can be with the following whole process of pseudo-representation:

R _i＝R _n

for?i＝1:N，

e _i＝[1，1，...，1]′/sqrt(M)；

for?k＝1：N _i，

y＝R _i＊e _i；

$e_i=\frac{y}{{\left|\right|y\left|\right|}_2};$

$a_i=e_i^T*R_i*e_i/(e_i^T*e_i);$

end，

$R_i=R_i-a_i*e_i*e_i^T;$

end

The approximate up space-time unite detection method of the Low-Rank of the invention described above is fit to fixed-point computation very much, has good numerical computations precision under limited dynamic range.Specifically in step 203, during the principal eigenvector of the iterative computation matrix of power method, can utilize simple shifting function to replace the normalized operation of vector; And see according to fixed-point simulation, in the time of each iterative computation dominant eigenvalue, because characteristic vector is near normalization, calculating that can contraction in division.Like this, whole iterative process is made up of matrix multiplication and the normalized like this operation of vector displacement substantially.In step 205, because the approximate subspace of the Low-Rank of spatial interference cross-correlation matrix is the low-dimensional vector space at the tenth of the twelve Earthly Branches that normalized vector is formed, though the process of matrix inversion is arranged,, lower dimension also makes the calculating of whole steps not have the possibility of big conditional number.

According to the detection method of the invention described above, carried out system emulation.Fig. 3 is under the N frequency point networking, according to the present invention with the performance that obtains of the signal detecting method of background technology schematic diagram relatively.The cumulative probability distribution (cumulative distribution) of demodulation signal interference ratio when as shown in Figure 3, curve 1 and curve 2 carry out input for using method of the present invention; The cumulative distribution of curve 3 demodulation signal interference ratio during for the input of no space prewhitening.Wherein, curve 1 is the MMSE-BLE result of the approximate prewhitening of Rank-1; Curve 2 is the MMSE-BLE result of the approximate prewhitening of Rank-N, and wherein, N is the reception antenna number.

As seen from Figure 3, method of the present invention adopts and also can obtain certain inhibition gain to co-channel interference under the simplest implementation.And by the embodiment of the invention described above as seen, signal detecting method of the present invention calculates simple relatively, and is very easy to, and is suitable for adopting fixed-point computation.In the method for the invention, because the effect of power control and channel estimating reprocessing, the component of useful signal is revealed the contribution of the estimation of the space association correlation matrix of interference signal limited, this method adopts the Low-Rank sub-space approximation to the estimation of the space association correlation matrix of interference signal, can not carry out albefaction to this sub-district echo signal and suppress.

Being preferred embodiment of the present invention only below, is not to be used to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of being done, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (8)

1, a kind of method for detecting space-time unite signal is characterized in that, this method comprises:

Determine spatial interference cross-correlation matrix R at receiving terminal by channel estimating _s

Calculate described spatial interference cross-correlation matrix R _sRank-N approximate matrix E, and determine spatial interference cross-correlation matrix R according to this approximate matrix E _sOrthogonal intersection cast shadow matrix

, with this orthogonal intersection cast shadow matrix As spatial interference cross-correlation matrix R _sInverse matrix, send the detection of signal, described N is a positive integer.

2, method according to claim 1 is characterized in that, described computer memory disturbs cross-correlation matrix R _nRank-N approximate matrix E comprise:

Utilize power method and progressively the null method computer memory disturb cross-correlation matrix R _sN eigenvalue of maximum, and calculate N corresponding respectively principal eigenvector e of this N characteristic value _i, wherein, the value of i is 1 to N;

Utilize described N principal eigenvector structure approximate matrix E=[e _i] _I=1:N

3, method according to claim 2 is characterized in that, described utilize power method and progressively the null method computer memory disturb cross-correlation matrix R _sN eigenvalue of maximum, and calculate N corresponding respectively principal eigenvector e of this N characteristic value _iComprise:

A, with spatial interference cross-correlation matrix R _sAs current matrix;

B, utilize characteristic value of current matrix computations and corresponding principal eigenvector thereof, and upgrade current matrix;

Whether the characteristic value number that c, judgement obtain is N, if, then continue to carry out the described operation that utilizes N principal eigenvector structure approximate matrix, otherwise, step b returned.

4, method according to claim 3 is characterized in that, describedly utilizes characteristic value of current matrix computations and corresponding principal eigenvector thereof, and upgrades current matrix and comprise:

B1, make that current principal eigenvector is e _i=[1,1 ..., 1] ^T/ sqrt (M), and wherein [1,1 ..., 1] be M dimension unit vector, M is the reception antenna number;

B2, the current principal eigenvector of renewal are $e_i=\frac{R_i*e_i}{{\left|\right|R_i*e_i\left|\right|}_2},$ Wherein, ‖ ‖ ₂Be 2 rank norms;

B3, according to current matrix and the current principal eigenvector after upgrading calculate current characteristic value and be $a_i={e_i}^T*R_i*e_i/({e_i}^T*e_i);$

B4, execution in step b2～b3 finishes iteration one time successively, up to reaching default iterations, carries out next step;

B5, the current characteristic value that obtains according to step b4 and current principal eigenvector upgrade current matrix and are $R_i=R_i-a_i*e_i*e_i^T.$

5, method according to claim 1 is characterized in that, describedly determines spatial interference cross-correlation matrix R according to this approximate matrix E _sOrthogonal intersection cast shadow matrix

Comprise:

Wherein, E ^TTransposed matrix for E.

6, method according to claim 1 is characterized in that, determines N according to the signal to noise ratio requirement of computation complexity, system resource and demodulation.

7, method according to claim 1 is characterized in that, describedly determines spatial interference cross-correlation R at receiving terminal by channel estimating _sComprise:

From the subscriber signal that receives, extract the data of training sequence part, utilize this data estimation original user up-converter channel impact responding h _i

According to interference power Piip and reprocessing thresholding threshold_pp, carry out the noise reduction reprocessing, obtain interference/noise channel weight matrix T;

According to described interference/noise channel weight matrix T, calculate the spatial interference cross-correlation matrix R of up reception _s

8, method according to claim 7 is characterized in that, the spatial interference cross-correlation matrix R of the up reception of described calculating _sComprise:

$R_s=\frac{\mathrm{PT}\&CenterDot;T^H}{(P-P1)},$ Wherein, P is the code length of described training sequence, and P1 is the active path number that described noise reduction reprocessing obtains.

Images