CN101431354B - Direction of arrival estimation method - Google Patents

Abstract

The invention discloses a method for direction of arrival estimation which is applied to a wireless communication system using a smart antenna, and is characterized in that after a channel impulse response matrix H of a user is obtained, conjugation rearrangement is carried out on the H or a complex conjugation matrix R* of a space covariance matrix R corresponding to the H to obtain a space covariance matrix R' which is in the conjugation rearrangement, and spectrum estimation is carried out on Rm equal to alpha mulitplies R' plus beta mulitplies R to obtain a direction of arrival angle estimated value. Compared with the existing DOA estimation method based on the space covariance matrix to estimate the spectrum, the DOA estimation method based on the space covariance matrix after the conjugation rearrangement to estimate the spectrum can efficiently enhance the accuracy of the DOA estimation, lowers the correlation among multipath signals of the user, and can improve the performance of carrying out the DOA estimation on the multipath signals of the desired user by utilizing high resolution algorithm in a system. Under the condition that the method does not reduce related multipath number which can be estimated, the calculation amount has no obvious increase.

Description

A kind of direction of arrival estimation method

Technical field

The present invention relates to a kind of ripple and reach angle (Direction ofArrival is called for short DOA) method of estimation.

Background technology

As the key technology in TD-SCDMA (Time Division-Synchronous Code Division Multiple Access) system, smart antenna has obtained more and more deep research in recent years.Signal processing technology was enlarging cell range, is improving power system capacity, is improving the system spectrum utilance, reducing transmitting power and is reducing the aspect such as user's interference to have shown huge potential when smart antenna combined form empty with traditional signal procesing in time domain technology.DOA estimates it is the basis that intelligent antenna wireless technological orientation and channel allocation realize.

For the DOA method of estimation, the simplest implementation method is that spectrum is estimated, namely utilize array format space wave to be reached the power azimuth spectrum of signal, such as Bartlett (Robin Bartlett) spectrum, Capon (card friend) spectrum, MUSIC (multiple signal classification) spectrum and ESPRIT (estimation of invariable rotary signal parameter) spectrum etc. estimated, and then obtained the Bo Dajiao of subscriber signal.

In numerous DOA methods of estimation, the MUSIC algorithm based on signal source covariance matrix feature decomposition that Schmidt (Schmidt) proposes is the most representative.This algorithm has been broken through the Rayleigh limit by the restriction of array aperture size, is a kind of high resolution DOA estimation method that is widely used.But the unrelated signal source only can be estimated at a distance of nearer signal in MUSIC algorithm indistinguishable space under low signal-to-noise ratio.For Correlated Signals, the MUSIC algorithm will be along with the increase of correlation between signal source (coefficient correlation), and performance runs down, until complete failure.And in the radio signal propagation environment, wireless channel is very complicated, from simple line-of-sight propagation to communication environments such as the reflection with various barriers, refraction and multipaths, there is in a large number coherent signal, need a kind of method can reduce correlation between the signal for this reason, reach the DOA method of estimation at angle with the ripple that correctly estimates the coherent multipath signal.

At document: Kundu D.Modify MUSIC Algorithm for Estimating DOA ofSignal[J] .Signal Processing, 1996, (48): 85～89 (" the MUSIC algorithms that are used for the DOA estimation improvement of signal ", " signal is processed, 1996 " the 85th page～89 pages, female all D.) in, the method for a kind of employing Search Space Smoothing (spatial smoothing technique) decorrelation (decorrelating) has been proposed.The method is that receiving array is divided into some submatrixs, obtains respectively the correlation matrix of each submatrix, then these correlation matrixes is averaged and obtains level and smooth correlation matrix, realizes decorrelation.Adopt the method for this space smoothing to carry out decorrelation and reduced array number and array aperture, and then reduced to estimate number of source, for forward direction space smoothing and front-rear space smooth, estimable information source number is respectively M/2 and 2M/3 (M is the base station array antenna number), and the DOA estimated accuracy to Correlated Signals is also lower simultaneously.

Summary of the invention

Technical problem to be solved by this invention is, overcomes the deficiencies in the prior art, proposes a kind ofly not reduce the DOA estimated accuracy, and do not reduce the DOA method of estimation of the number that can estimate multipath signal.

In order to address the above problem, the invention provides a kind of direction of arrival estimation method, be applied to use in the wireless telecommunication system of smart antenna, it is characterized in that, after obtaining user's channel impulse response matrix H, to the complex-conjugate matrix R of space covariance matrix R corresponding to H or H ^*Carry out conjugation and reset, obtain the space covariance matrix R ' that conjugation is reset, to R _m=α * R '+β * R composes and estimates to obtain direction of arrival angle estimation value; Wherein, α＞0, β 〉=0.

In addition, adopt following method to carry out conjugation to described H and reset, obtain the channel impulse response matrix H that conjugation is reset _m: H _m=I * H ^*And according to H _mAdopt following method to obtain described R ': R '=H _m* (H _m) ^HWherein, above-mentioned I is and H ^*The reverse unit matrix of same order, (H _m) ^HBe H _mAssociate matrix, H ^*Complex-conjugate matrix for H.

In addition, adopt following method to described R ^*Carry out conjugation and reset and obtain described R ': R '=I * R ^** I; Above-mentioned I is and R ^*The reverse unit matrix of same order.

In addition, described α=β=0.5, or α=β=1, or α=1, β=0.

In addition, adopt emulation mode to obtain the value of described α, β, so that when carrying out described spectrum estimation, use α, described R corresponding to β value _mThe spatial spectrum function of structure satisfies required precision.

In addition, described spectrum is estimated as: Bartlett composes estimation, and Capon composes estimation, and MUSIC composes estimation, one of ESPRIT spectrum.

In addition, described wireless telecommunication system is the TD-SCDMA system.

Compare with existing DOA method of estimation of composing estimation based on space covariance matrix, adopt above-mentioned DOA method of estimation of composing estimation based on the space covariance matrix after the conjugation rearrangement of the present invention, can effectively improve the precision that DOA estimates, reduce the correlation between the user multipaths signal, can utilize high resolution algorithm to the DOA estimated performance of desired user multipath signal in the improvement system.And method of the present invention is not in the situation that not reduce and can estimate the coherent multipath number, and amount of calculation there is no obvious increase.

Description of drawings

Fig. 1 is the direction of arrival estimation method flow chart that the embodiment of the invention is applied to the TD-SCDMA system.

Embodiment

Basic ideas of the present invention are that the channel impulse response matrix to the user carries out conjugation and resets, and the channel impulse response matrix that uses conjugation to reset generates the space covariance matrix of revising, and carries out DOA based on the space covariance matrix of above-mentioned correction and estimate.

The space smoothing method is a kind of a kind of effective ways that reduce signal correlation, its basic thought is that uniform line-array is divided into several overlapped subarrays, the array manifold of each submatrix is identical, the covariance matrix of each submatrix can average computing, realize decorrelation, this is the level and smooth basic principle of forward direction.Before and after to smoothly be except forward direction smoothly, each submatrix is carried out conjugation more oppositely reconstitute a subarray, then respectively each is averaged computing through front and back to the covariance matrix of the submatrix of level and smooth reprocessing, consist of total covariance matrix, realize decorrelation.

If it is identical with array element number to get submatrix length, just can reach and not dwindle array aperture, in the situation that do not affect algorithm unrelated signal DOA is estimated, improve the DOA estimated performance to coherent signal.The present invention just is being based on this thought the channel impulse response Matrix Conjugate is being reset, and realization is level and smooth to space covariance matrix, reaches the purpose that reduces the multipath signal correlation.

Describe the present invention below in conjunction with drawings and Examples.

Fig. 1 is the direction of arrival estimation method flow chart that the embodiment of the invention is applied to the TD-SCDMA system.As shown in Figure 1, the method comprises following steps:

S101: the base station obtains each user's aerial array channel impulse response matrix according to the channel estimating of each user on each antenna;

The mathematic(al) representation of the wireless channel model of using in the TD-SCDMA system is:

$h_k(\mathrm{\τ},\mathrm{\φ})=\underset{l}{\overset{L}{\mathrm{\Σ}}}{g}_{k,l}\left(t\right)\mathrm{\δ}(\mathrm{\τ}-{\mathrm{\τ}}_{k,l})\mathrm{\δ}\left({\mathrm{\φ}-\mathrm{\φ}}_{k,l}\right)$

Wherein, h _kK user's of (τ, φ) expression channel response, it is the function of time delay τ and angle of arrival φ.L represents the number of path that channel is total, and l represents each multipath label, g _{K, l}(t), τ _{K, l}, φ _{K, l}The time varying signal amplitude, relative time delay and the relative angle of arrival that represent respectively k user's l bar multipath.

In the situation that array antenna is used in the base station, when the signal that UE (user terminal) sends arrived the base station through multipath channel, the serve as reasons directivity channel impulse response of different arrival bearings (DOA) incident of the channel impulse response on each reception antenna was formed by stacking.If L _kRepresent to have when signal that k user sends incides base station array antenna different arrival bearings' (DOA) incoming signal number, observe from the time domain angle, if the discrete tap number of the channel impulse response after the stack is W, then channel impulse response (CIR) can be expressed as W * 1 dimension dispersion vector, i.e. L on the time domain _kThe stack of individual independently multipath component is actually L _kThe weighted superposition of individual CIR vector correspondence position element, weights i.e. the amplitude of fading of this direction.

Note h ^{(k, l)}=[h ^{(k, l)}(1), h ^{(k, l)}(2) ... h ^{(k, l)}(W)] ^TIt is the directivity channel impulse response vector of l incoming signal of k user; K all directivity channel impulse response vector of user can form a directivity channel impulse response matrix:

$H^{\left(k\right)}=\left[\begin{array}{c}{h}^{(k,1)}\left(1\right),h^{(k,1)}\left(2\right),...,h^{(k,1)}\left(W\right)\\ h^{(k,2)}\left(1\right),h^{(k,2)}\left(2\right),...,h^{(k,2)}\left(W\right)\\ \·\·\·\·\·\·\\ h^{(k,M)}\left(1\right),h^{(k,M)}\left(2\right),...,h^{(k,M)}\left(W\right)\end{array}\right];$

Wherein, M represents the base station array antenna number, and W represents channel estimation window long (W=128).

S102: according to user's channel impulse response matrix computations space covariance matrix;

K user's space covariance matrix R ^(k)For:

R ^(k)＝H ^(k)×(H ^(k)) ^H；

Wherein, (H ^(k)) ^HRepresenting matrix H ^(k)Associate matrix.

S103: the channel impulse response matrix to the user carries out the conjugation rearrangement;

According to the meaning of matrix multiplication as can be known, realize the rearrangement of array, during the unit of getting inverted matrix, the receiving terminal data of pair array need to be carried out premultiplication.Therefore, when the channel impulse response data are carried out the conjugation rearrangement, need to be to the reverse unit matrix of conjugation premultiplication of channel impulse response matrix.

Make that I is M * reverse unit matrix in M rank, that is:

$I={\left[\begin{array}{cccc}0& \·\·\·& 0& 1\\ 0& & 1& 0\\ & \·\·\·& \·\·\·& \\ 1& 0& \·\·\·& 0\end{array}\right]}_{M\×M};$

Channel impulse response (CIR) matrix to k user carries out the conjugation rearrangement, obtains the channel impulse response matrix H that conjugation is reset _Modify1 ^(k):

H _modify1 ^(k)＝I×H ^(k)*；

Wherein, H ^{(k) *}Be H ^(k)Complex-conjugate matrix.

S104: the space covariance matrix of calculating the channel impulse response matrix after above-mentioned conjugation is reset;

The channel impulse response matrix H that k user's conjugation is reset _Modify1 ^(k)Space covariance matrix be R _Modify1 ^(k):

R _modify1 ^(k)＝H _modify1 ^(k)×(H _modify1 ^(k)) ^H＝I×R ^(k)*×I；

S105: based on user's space covariance matrix and the space covariance matrix after the conjugation rearrangement, the space covariance matrix R that obtains revising _Modify ^(k)

In the present embodiment, get R _Modify ^(k)=(R _Modify1 ^(k)+ R ^(k))/2, the space covariance matrix of namely revising are covariance matrix average after space covariance matrix and conjugation are reset.

In another embodiment, can get:

R _Modify ^(k)=α * R _Modify1 ^(k)+ β * R ^(k)(formula 1)

α＞0 wherein, β 〉=0; The space covariance matrix of namely revising is the weighted average of the covariance matrix after space covariance matrix and conjugation are reset, and when β=0, the space covariance matrix of correction is the covariance matrix after the conjugation rearrangement.

The span of above-mentioned α and β obtains by emulation usually, the R that certainly obtains according to formula 1 _Modify ^(k)Need to satisfy the requirement that follow-up spectrum estimated matrix decomposes.

Can draw through emulation, when α=β=0.5, or α=β=1, or α=1, β=0 o'clock adopts above-mentioned formula 1 can obtain the DOA estimated result of same accuracy.

S106: based on the space covariance matrix R of above-mentioned correction _Modify ^(k)Compose estimation, the multipath arrival bearing of signal is estimated, obtain multipath direction of arrival angle;

In implementing use-case, adopt MUSIC spectrum method of estimation, construct namely that the multipath arrival bearing to signal estimates after the MUSIC spectral function.

By the space covariance matrix R of MUSIC Estimation of Spatial Spectrum to k user's correction _Modify ^(k)Carry out feature decomposition or SVD (singular value) decomposition or QR (QR) and decompose, computation of characteristic values and characteristic vector can get:

${R_{\mathrm{mod}\mathrm{ify}}}^{\left(k\right)}=U_S{\mathrm{\Σ}}_S{U}_S^H+U_N{\mathrm{\Σ}}_N{U}_N^H;$

Wherein, U _SThat the subspace of being opened by large characteristic value characteristic of correspondence vector is signal subspace, U _NThat the subspace of being opened by little characteristic value characteristic of correspondence vector is noise subspace.

Signal subspace under the ideal conditions in the data space and noise subspace are mutually orthogonal, i.e. steering vector a in the signal space ^H(θ) also with noise subspace quadrature, i.e. a ^H(θ) U _N=0.

The MUSIC spectral function of structure space covariance matrix:

$P_{\mathrm{MUSIC}}\left(\mathrm{\θ}\right)=\frac{1}{a^H\left(\mathrm{\θ}\right)U_N{U_N}^H\left(\mathrm{\θ}\right)a\left(\mathrm{\θ}\right)};$

The MUSIC spectrum peak is searched for, obtained multipath direction of arrival angle corresponding to user.

Spatial spectrum function P by structure _MUSIC(θ), when denominator is 0, P _MUSIC(θ) spike is arranged, θ is changed, estimate the angle of arrival DOA of multipath signal by the search spectrum peak, with this high resolution DOA estimation that realizes the space covariance matrix that obtains based on the channel impulse response Conjugate Data Rearrangement, realize that ripple to the multipath signal of desired user reaches the angle and estimates.

As from the foregoing, the present invention is in the channel impulse response matrix H that obtains the user ^(k)After, according to H ^(k)Obtain the covariance matrix R that conjugation is reset _Modify1 ^(k)According to R _Modify1 ^(k)Carrying out follow-up spectrum estimates.And above-mentioned R _Modify1 ^(k)Can adopt dual mode to obtain:

(1) to H ^(k)Carry out the conjugation rearrangement and obtain H _Modify1 ^(k), and according to H _Modify1 ^(k)Calculate R _Modify1 ^(k)

(2) use H ^(k)Calculate R ^(k), and to R ^{(k) *}Carry out conjugation and reset acquisition R _Modify1 ^(k)

The purpose of above dual mode all is to reset by conjugation to obtain the little covariance matrix of correlation, and composes estimation according to this covariance matrix, obtains the multipath direction of arrival angle estimation value of degree of precision with the cost of minimum.

Compare with existing DOA method of estimation of composing estimation based on space covariance matrix, adopt above-mentioned DOA method of estimation of composing estimation based on the space covariance matrix after the conjugation rearrangement of the present invention, can effectively improve the precision that DOA estimates, reduce the correlation between the user multipaths signal, can utilize high resolution algorithm to the DOA estimated performance of desired user multipath signal in the improvement system.And method of the present invention is not in the situation that not reduce and can estimate the coherent multipath number, and amount of calculation there is no obvious increase, only increases M when the computer memory covariance matrix ²Sub-addition.

MUSIC spectrum in above-described embodiment was estimated, the DOA method of estimation based on the space covariance matrix after the conjugation rearrangement of the present invention also can adopt other any spectrum method of estimation based on space covariance matrix.

In addition, above-described embodiment is only introduced DOA method of estimation of the present invention as an example of the TD-SCDMA system example, in addition, it is different that the system of the employing smart antenna that also is applicable to other standard of the space covariance matrix after resetting based on conjugation of the present invention, difference only are that the base station obtains each user's the method for aerial array channel impulse response matrix.

Claims (7)

1. direction of arrival estimation method, be applied to use in the wireless telecommunication system of smart antenna, it is characterized in that, after obtaining user's channel impulse response matrix H, the complex-conjugate matrix R* of space covariance matrix R corresponding to H or H is carried out conjugation to be reset, obtain the space covariance matrix R ' that conjugation is reset, to the space covariance matrix R that revises _m=α * R '+β * R composes and estimates to obtain direction of arrival angle estimation value; Wherein, the space covariance matrix of correction is the weighted average of the covariance matrix after space covariance matrix and conjugation are reset, α＞0, β 〉=0.

2. direction of arrival estimation method as claimed in claim 1 is characterized in that, adopts following method to carry out conjugation to described H and resets, obtain the channel impulse response matrix H that conjugation is reset _m: H _m=I * H ^*And according to H _mAdopt following method to obtain described R ': R '=H _m* (H _m) ^HWherein, above-mentioned I is and H ^*The reverse unit matrix of same order, (H _m) ^HBe H _mAssociate matrix, H ^*Complex-conjugate matrix for H.

3. direction of arrival estimation method as claimed in claim 1 is characterized in that, adopts following method that described R* is carried out conjugation and resets and obtain described R ': R '=I * R* * I; Above-mentioned I is the reverse unit matrix with the R* same order.

4. direction of arrival estimation method as claimed in claim 1 is characterized in that, described α=β=0.5, or α=β=1, or α=1, β=0.

5. direction of arrival estimation method as claimed in claim 1 is characterized in that, adopts emulation mode to obtain the value of described α, β, so that when carrying out described spectrum estimation, use α, described R corresponding to β value _mThe spatial spectrum function of structure satisfies required precision.

6. direction of arrival estimation method as claimed in claim 1 is characterized in that, described spectrum is estimated as: Bartlett composes estimation, and Capon composes estimation, and MUSIC composes estimation, one of ESPRIT spectrum.

7. direction of arrival estimation method as claimed in claim 1 is characterized in that, described wireless telecommunication system is the TD-SCDMA system.

Images