CN101667860B - Method for detecting signals in multi-antenna digital wireless communication system - Google Patents

Abstract

The embodiment of the invention provides a method for detecting signals in a multi-antenna digital wireless communication system. The method comprises that: at least two receiving antennas of a receiving end receive transmitting signals, and acquire at least two receiving signals; the receiving end performs channel estimation according to the receiving signals to acquire a channel matrix H consisting of channel coefficients between the transmitting antennas and the receiving antennas; an estimation error covariance matrix of partial transmitting signals in all the transmitting signals is calculated by using the channel matrix H, then newly increased matrix items are acquired by using the channel matrix H and the calculated estimation error covariance matrix of the partial transmitting signals, and further estimation error covariance matrixes of all the transmitting signals are recursively acquired; and all the transmitting signals are detected by using the acquired estimation error covariance matrixes of all the transmitting signals. The method provided by the embodiment of the invention reduces the calculation complexity of detecting the signals, the use of hardware resources, and the detection time delay.

Description

The method of signal is detected in multi-antenna digital wireless communication system

It was submitted this application claims on September 1st, 2008, application No. is CN200810141750.5, the priority of China's application of entitled " method of signal is detected in multi-antenna digital wireless communication system ", entire contents are hereby incorporated by reference in the application.

Technical field

The present invention relates to signal detection techniques, particularly relate to a kind of method that signal is detected in multi-antenna digital wireless communication system.

Background technique

According to information theory, use multi-antenna array that can greatly improve transmission bit rate simultaneously in the transmitting terminal of communication system and receiving end.

Use the wireless communication system with Space-Time framework of multi-antenna array as shown in Figure 1 simultaneously in transmitting terminal and receiving end.System work approximate can regard statistical iteration as in Rayleigh scattering environment, each element of channel matrix.In the system shown in figure 1, a data sequence is divided into M incoherent symbol subsequences, and each subsequence is emitted by one of M transmitting antenna.M subsequence is received in receiving end by N number of receiving antenna after the influence for the channel that a channel matrix is H.Emit signal s₁..., s_MIt is corresponding to receive signal x respectively by M different antenna element a-1 ..., a-M transmittings₁..., x_NIt is received respectively from N number of different antenna element b-1 ..., b-N.In the system, transmission antenna unit number M is at least 2, and receiving antenna unit number N is at least M.Channel matrix H is the matrix of a N × M, the coupling that i-th of receiving antenna of element representation and j-th of transmitting antenna that the i-th row j is arranged in matrix pass through transmission channel.Receive signal x₁..., x_NThe transmitting signal of recovery is treated to generate in digital signal processorSummation ingredient c-1, c-2 ..., c-N are also shown in this figure, they represent the unavoidable noise signal w for including₁, w₂..., w_N, these noise signals are added separately in the signal that receiving antenna unit b-1, b-2 ..., b-N are received.

It is channel matrix by the matrix that the channel coefficients between transmitting antenna and receiving antenna form, channel coefficients are to carry out channel estimation using reception signal to obtain.Channel matrix H in the system shown in figure 1 is the matrix of a N × M, is indicated are as follows:

Channel matrix H is N × M complex matrix, it is assumed that the channel matrix in the period of K symbol in be constant.Vector h_N:(n=1,2 ..., N) and h_{: m}The length of (m=1,2 ..., M) is M and N respectively.Wherein, the channel vector h that channel matrix H includes_{: 1}To h_{: M}Respectively indicate the influence that channel transmits each transmission signal in signal to M.It is more specific, channel vector h_{: m}(m=1,2 ..., M) includes channel matrix entry h_1mTo h_Nm, receiving antenna unit b-1 is illustrated respectively in into b-N on each receiving antenna, and channel is to transmitting signal s_mInfluence.

In the system shown in figure 1, emit and meet relational expression between the vector of signal and the vector of reception signal $x\left(k\right)=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{h}_{:m}{s}_m\left(k\right)+w\left(k\right)=\mathrm{Hs}\left(k\right)+w,$ Wherein k indicates sampling instant, k=1,2 ..., K.Indicate that above-mentioned relation is with vector form $\stackrel{\&RightArrow;}{x}=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{h}_{:m}{s}_m+\stackrel{\&RightArrow;}{w}=H\stackrel{\&RightArrow;}{s}+\stackrel{\&RightArrow;}{w},$ The formula is written as again $\stackrel{\&RightArrow;}{x}={\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">s</figure-callout>}}_1\&CenterDot;h_{:1}+{\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">s</figure-callout>}}_2\&CenterDot;h_{:2}+...+s_m\&CenterDot;h_{:m}+\&CenterDot;\&CenterDot;\&CenterDot;+s_M\&CenterDot;h_{:M}+\stackrel{\&RightArrow;}{w}$ Form, can be clearly seen that each transmitting signal to received signal vector

Influence.

The least mean-square error (MMSE) of transmitting signal is estimated as $\widehat{\stackrel{\&RightArrow;}{s}}={(H^H\&CenterDot;H+\mathrm{\&alpha;}{I}_{M\&times;M})}^{-1}{H}^H\stackrel{\&RightArrow;}{x},$ Wherein, symbol^-1Inverse of a matrix matrix is sought in expression, and α is constant relevant to the transmitting signal-to-noise ratio of signal, $\mathrm{\&alpha;}=\frac{{\mathrm{\&sigma;}}_{\mathrm{<figure-callout\; id="2"\; label="w"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">w</figure-callout>}}^2}{{\mathrm{\&sigma;}}_s^2}.$

Evaluated error of the present invention $e=\stackrel{\&RightArrow;}{s}-\widehat{\stackrel{\&RightArrow;}{s}}$ Covariance matrix be the normalized square mean of additive white Gaussian noise be 1 in the case where covariance matrix, i.e., $E\left\{(\stackrel{\&RightArrow;}{s}-\widehat{\stackrel{\&RightArrow;}{s}}){(\stackrel{\&RightArrow;}{s}-\widehat{\stackrel{\&RightArrow;}{s}})}^H\right\}={(H^H\&CenterDot;H+\mathrm{\&alpha;}{I}_{M\&times;M})}^{-1},$ It is denoted as Q, and defines R=(H^H·H+αI_M×M), then there is Q=R^-1.To be expressed as to the estimated value of transmitting signal $\widehat{\stackrel{\&RightArrow;}{s}}={\mathrm{QH}}^H\stackrel{\&RightArrow;}{x}$ Form.

In the prior art, in above-described MIMO (multiple-input and multiple-output, Multiple-inputMultiple-output the method that signal detection) is realized in system is: the initial value of the evaluated error covariance matrix Q of all transmitting signals to be detected is obtained first with channel matrix H, then using the calculation of initial value of obtained Q to the estimated value of transmitting signal.Wherein, the initial value for how calculating Q will will affect the calculation amount and complexity of signal detection.

In the prior art, the initial value of Q is obtained by following recurrence method:

The sequencing that all M transmitting signals are detected in receiving end is preset, is denoted as t with transmitting signal serial number_M, t_M-1..., t_m..., t₂, t₁；Correspondingly, channel matrix H is obtained by column rearrangement $H_M^{\left(t_M\right)}=[h_{:t_1}{h}_{{:\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}\&CenterDot;\&CenterDot;\&CenterDot;h_{:t_{M-1}}{h}_{{:t}_M}],$ Wherein

Indicate the t of channel matrix H_mColumn.Estimation error covariance matrix Q is acquired again_MInverse matrix $R_M={\left(H_M^{\left(t_M\right)}\right)}^H\&CenterDot;H_M^{\left(t_M\right)}+\mathrm{\&alpha;}{I}_{M\&times;M}.$ Corresponding finally detected m (m=1,2 ..., M) a transmitting signal t_m..., t₂, t₁Evaluated error covariance matrix inverse matrix $R_m^{\left(t_m\right)}={\left(H_m^{\left(t_m\right)}\right)}^H\&CenterDot;H_m^{\left(t_m\right)}+\mathrm{\&alpha;}{I}_{\left(m\right)\&times;\left(m\right)},$ AndWith

There is following recurrence relation: $R_m^{\left(t_m\right)}=\left[\begin{array}{cc}{R}_{m-1}^{\left(t_{m-1}\right)}& v_{m-1}^{\left(t_m\right)}\\ {\left(v_{m-1}^{\left(t_m\right)}\right)}^H& {\mathrm{\&beta;}}_1^{\left(t_m\right)}\end{array}\right],$ Wherein,It is the m-1 transmitting signal t that correspondence that last recursion obtains finally is detected_m-1..., t₂, t₁Evaluated error covariance matrix inverse matrix, ${\mathrm{\&beta;}}_1^{\left(t_m\right)}={h_{{:t}_m}}^H\&CenterDot;h_{{:t}_m}+\mathrm{\&alpha;},$ $v_{m-1}^{\left(t_m\right)}=\left[\begin{array}{c}{h_{{:t}_1}}^H\&CenterDot;h_{{:t}_m}\\ {h_{{:t}_1}}^H\&CenterDot;h_{{:t}_m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {h_{{:t}_{m-1}}}^H\&CenterDot;h_{{:t}_m}\end{array}\right].$

Calculate a finally detected transmitting signal t₁Corresponding evaluated error covariance matrix, is denoted as $Q_1^{\left(t_1\right)}={\left(R_1^{\left(t_1\right)}\right)}^{-1}.$

Utilize the evaluated error covariance matrix of a corresponding finally detected transmitting signal

Or the m-1 transmitting signal t that the obtained correspondence of last recursion is finally detected_m-1..., t₂, t₁Evaluated error covariance matrix

Recursion obtains

Recurrence method is as described below:

It calculates first

Thank to Germania-Morrison (Sherman-Morrison) as a result, i.e. use is thanked to Germania-Morrison (Sherman-Morrison) formula and obtained

$T_{m-1}^{\left(t_m\right)}=Q_{m-1}^{\left(t_{m-1}\right)}+\frac{Q_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{Q}_{m-1}^{\left(t_{m-1}\right)}}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}-{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{Q}_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}};$ Then by

And

Obtain m detected to the end transmitting signal t_m..., t₂, t₁Corresponding evaluated error covariance matrix

For, $Q_M^{\left(t_m\right)}=\left[\begin{array}{cc}{T}_{m-1}^{\left(t_m\right)}& -\frac{T_{m-1}^{\left(t_m\right)}{v}_{m-1}^{\left(t_m\right)}}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}}\\ -{\left(\frac{T_{m-1}^{\left(t_m\right)}{v}_{m-1}^{\left(t_m\right)}}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}}\right)}^H& \frac{1}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}}+\frac{{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{T}_{m-1}^{\left(t_m\right)}{v}_{m-1}^{\left(t_m\right)}}{{\left({\mathrm{\&beta;}}_1^{\left(t_m\right)}\right)}^2}\end{array}\right].$

Using it is above-mentioned by

It obtains

Recurrence method, successively by

Recursion

Again by

RecursionSo until recursion obtains all M transmitting signal t_M, t_M-1..., t_m..., t₂, t₁Corresponding evaluated error covariance matrix

Value.

It is exactly optimal detection sequence to be determined in signal detection process, and when gradually detecting according to the optimal detection sequence and using interference elimination method each transmitting signal, which is denoted as Q by the initial value of used evaluated error covariance matrix_M, $Q_M=Q_M^{\left(t_M\right)}.$

In the prior art, the method after the initial value for obtaining Q in aforementioned manners, using the calculation of initial value of Q to the estimated value for emitting signal are as follows:

Assuming that m is the iteration variable detected in signal process, then the corresponding evaluated error covariance matrix of the transmitting signal to be detected of M+1-m when detecting m-th of transmitting signal is denoted as Q_M+1-m。

As m=1, i.e., when detecting first transmitting signal, Q at this time_M+1-mFor the initial value of Q, therefore first estimated value for being detected transmitting signal directly is obtained using the calculation of initial value of Q.Then, the corresponding evaluated error covariance matrix Q of remaining M-1 transmitting signal to be detected is obtained using Q recursion_M-1。

As m > 1, the corresponding evaluated error covariance matrix Q of M+1-m transmitting signal to be detected is corresponded at this time_M+1-mIt is calculated after upper primary detection transmitting signal, and utilizes Q_M+1-mM-th of estimated value for being detected transmitting signal is calculated.Then, Q is utilized_M+1-mRecursion obtains the corresponding evaluated error covariance matrix Q of remaining M-m transmitting signal to be detected_M-m。

And utilize matrix Q_M+1-mRecursion obtains the corresponding evaluated error covariance matrix Q of remaining M-m transmitting signal to be detected_M-mMethod it is as follows:

From matrix Q_M+1-mThe matrix that middle deletion M+1-m row and M+1-m are arranged is denoted as T_M-m, matrix Q_M+1-mIn M+1-m column first M-m form column vectors be denoted as w_M-m, and matrix Q_M+1-mIn be located at the item that M+1-m row and M+1-m are arranged and be denoted as ω_M-m, i.e., $Q_{M+1-m}=\left[\begin{array}{cc}{T}_{M-m}& w_{M-m}\\ {\left(w_{M-m}\right)}^H& {\mathrm{\&omega;}}_{M-m}\end{array}\right].$ So, Q is calculated_M-mMethod be: using T_M-m、w_M-mAnd ω_M-mObtain Q_M-m, i.e., $Q_{M-1}=T_{M-m}-\frac{1}{{\mathrm{\&omega;}}_{M-m}}{w}_{M-m}\&CenterDot;{\left(w_{M-m}\right)}^H.$

On the other hand, Space-Time Block Coding (STBC) is a kind of space time coding scheme, and STBC utilizes the space diversity of signal, mimo system is enabled to obtain bigger channel capacity and signal gain.Alamouti scheme is a simple and classical example of STBC.In Alamouti Space-Time Block Coding technology, transmitting terminal uses two emitting antennas to emit signal simultaneously, or emits signal using more than two transmitting antennas simultaneously.One or more receiving antenna can be used in receiving end and receive signal.Two or more transmitting antennas are used simultaneously in transmitting terminal, the diversity gain of the two transmitting antennas can be obtained for receiving end.

Assuming that transmitting antenna number M=2, receiving antenna number N=2, the then signal that receiving end receives can be expressed as form:

$r=H\&CenterDot;a+v=\left[\begin{array}{cc}{h}_{11}& h_{12}\\ h_{21}& h_{22}\end{array}\right]\left[\begin{array}{cc}{a}_1& -a_2^{*}\\ a_2& a_1^{*}\end{array}\right]+\left[\begin{array}{cc}{v}_{11}& v_{12}\\ v_{21}& v_{22}\end{array}\right]$

Wherein, r is to receive signal, and H is channel matrix, and a is transmitting symbol, and v is noise.R, the definition of H, a are all across 2 symbol periods, 2 symbol periods, 2 symbol periods in referred to as 1 Alamouti Space-Time Block Coding period.In first symbol period, two emitting antennas emits a respectively₁And a₂；In second symbol period transmitting-a respectively^* ₂And a^* ₁；Channel coefficients in H remain unchanged in 2 symbol periods；Receiving antenna receives the symbol of transmitting terminal two emitting antennas transmitting respectively in two symbol periods, and the signal that i-th of receiving antenna receives in two symbol periods is respectively r_i1=h_i1a₁+h_i2a₂+v₁With ${\mathrm{<figure-callout\; id="2"\; label="r\; i"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">r</figure-callout>}}_i=-h_{i1}{a}_2^{*}+h_{i2}{a}_1^{*}+{\mathrm{<figure-callout\; id="2"\; label="v"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">v</figure-callout>}}_2.$

A symbol period as described herein refers to the section occupied in the time domain by a symbol of transmission, the section perhaps occupied on frequency domain or the section occupied on the two-dimensional surface of time domain and frequency domain.Such as, one data packet uses 8 OFDM symbols in time domain, each OFDM symbol occupies 16 subcarriers on frequency domain, so symbol period, just refer to a section on the two-dimensional surface of time domain and frequency domain, 1 subcarrier namely in time domain in 1 OFDM symbol, and this data packet shares 8 × 16=128 symbol period.

Further, it is also possible to be, transmitting terminal has 4 transmitting antennas, and there are at least two receiving antennas in receiving end.4 transmitting antennas of transmitting terminal are divided into two groups, every group of two emitting antennas.Two emitting antennas in group emits one group of Alamouti Space-Time Block Coding, and each group emits different symbols respectively.

For the above-mentioned MIMO communication system using Alamouti Space-Time Block Coding, there are a kind of methods of receiving end detection signal, but the receiver in this method realizes that technology generally requires the operation of extraction of square root, and during actual fixed-point implementation, the realization of extraction of square root operation is often extremely complex, to need to avoid as possible.

Present inventor during realization of the invention, discovery in the prior art the prior art has at least the following problems:

1) due in the prior art, using the prior art by

It obtains

Recurrence method, successively by

Recursion

Again by

Recursion

So until recursion obtains the corresponding evaluated error covariance matrix of all M transmitting signals

Value, the recursive process complexity of the initial value of the Q of calculating is higher, cause receive machine testing signal used in hardware resource it is more, and detect time delay it is larger.

2) prior art be directly generalized to by spatial reuse emit multichannel emit signal in comprising at least all the way the scene of Alamouti Space-Time Block Coding (STBC) when, calculation amount is larger, cause hardware resource used in reception machine testing signal more, and it is larger to detect time delay.

Summary of the invention

In view of this, the embodiment of the present invention proposes a kind of method for detecting signal in multi-antenna digital wireless communication system, the complexity for calculating Q initial value is reduced, that reduces hardware resource uses and reduce detection time delay.

A kind of method that signal is detected in multi-antenna digital wireless communication system that the embodiment of the present invention proposes, the at least two transmitting signal of detection in multiple-input, multiple-output mimo system, the transmitting signal is emitted respectively by each different transmitting antenna of transmitting terminal and reaches receiving end by a channel, this method comprises:

A. at least two receiving antennas of receiving end receive the transmitting signal, obtain at least two and receive signal；

B. receiving end carries out channel estimation according to signal is received, and obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna；

C. the evaluated error covariance matrix of the partial transmitting signal in all transmitting signals is calculated using channel matrix H, then the evaluated error covariance matrix of calculated partial transmitting signal and the newly-increased matrix entries obtained according to the evaluated error covariance matrix and channel matrix H of the partial transmitting signal are utilized, recursion is acquired including the partial transmitting signal and number is more than the evaluated error covariance matrix of the transmitting signal of the partial transmitting signal number, the newly-increased matrix entries be include the newly-increased matrix entries of the partial transmitting signal and number more than the evaluated error covariance matrix of the evaluated error covariance matrix partial transmitting signal of the transmitting signal of the partial transmitting signal number；

D. the obtained evaluated error covariance matrix of step c is utilized, includes the transmitting signal that partial transmitting signal and number are more than the partial transmitting signal number described in detecting step c.

The method of signal detection in another multi-antenna digital wireless communication system that the embodiment of the present invention proposes, at least two symbols of transmitting terminal transmitting are detected in multiple-input, multiple-output mimo system, wherein at least one symbol at least two symbol is emitted by least one transmitting antenna again after transmitting terminal carries out channel coding by an encoder；The encoder carries out channel coding to input symbol in such a way that symbol is duplicate and obtains channel signal, the channel signal includes a former input symbol, or the negative value of symbol is inputted including one, or the multiple conjugate value of symbol is inputted including one, or the negative multiple conjugate value of symbol is inputted including one, the channel signal of the encoder output, which is emitted by least one transmitting antenna and passes through at least two different channels, reaches receiving end；This method comprises:

A. at least two receiving antennas of receiving end receive the channel signal that transmitting terminal is emitted, and obtain at least two and receive signal；

B. receiving end carries out channel estimation, obtains the channel matrix being made of the channel coefficients between transmitting antenna and receiving antenna；

C. the evaluated error covariance matrix of the partial symbols at least two symbol is calculated using channel matrix, then the evaluated error covariance matrix of the partial symbols and the newly-increased matrix entries obtained according to the evaluated error covariance matrix and channel matrix H of the partial symbols are utilized, recursion is acquired including the partial symbols and number is more than the evaluated error covariance matrix of the symbol of the partial symbols number, the newly-increased matrix entries be include the newly-increased matrix entries of the partial transmitting signal and number more than the evaluated error covariance matrix of the evaluated error covariance matrix partial transmitting signal of the transmitting signal of the partial transmitting signal number；

D. some items in the obtained evaluated error covariance matrix of step c or evaluated error covariance matrix, at least one of at least two symbols of detection transmitting terminal transmitting are utilized.

The method that the embodiment of the present invention also proposed signal detection in a kind of multi-antenna digital wireless communication system, the at least two transmitting signal of detection in multiple-input, multiple-output mimo system, the transmitting signal is emitted respectively by each different transmitting antenna of transmitting terminal and reaches receiving end by a channel, this method includes

A. at least two receiving antennas of receiving end receive the transmitting signal, obtain at least two and receive signal；

B. receiving end carries out channel estimation according to signal is received, and obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna；

C. a sub- inverse of a matrix matrix of matrix H is calculated using channel matrix H, then a sub- inverse of a matrix matrix of calculated matrix H and the newly-increased matrix entries obtained according to channel matrix H are utilized, recursion acquires the inverse matrix of a submatrix comprising the H and a bigger submatrix of the H of a submatrix greater than the H, and the newly-increased matrix entries are the newly-increased matrix entries of a submatrix comprising the H and a submatrix of a bigger submatrix inverse matrix H of the H of a submatrix greater than the H；

D. using the inverse matrix of a bigger submatrix of the obtained H of step c, transmitting signal described in detecting step a.

The method that the embodiment of the present invention also proposed signal detection in a kind of multi-antenna digital wireless communication system, the at least two transmitting signal of detection in multiple-input, multiple-output mimo system, the transmitting signal is emitted respectively by each different transmitting antenna of transmitting terminal and reaches receiving end by a channel, this method includes

A. at least two receiving antennas of receiving end receive the transmitting signal, obtain at least two and receive signal；

B. receiving end carries out channel estimation according to signal is received, and obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna；

C. the LDU factoring matrix of a submatrix of H-matrix H is calculated using channel matrix H, then the LDU factoring matrix of a submatrix of calculated H and the newly-increased matrix entries obtained according to channel matrix H are utilized, recursion acquires the LDU factoring matrix of a submatrix comprising the H and a bigger submatrix of the H of a submatrix greater than the H, and the newly-increased matrix entries are the newly-increased matrix entries of a submatrix comprising the H and a submatrix of a bigger submatrix inverse matrix H of the H of a submatrix greater than the H；

D. using the LDU factoring matrix of a bigger submatrix of the obtained H of step c, transmitting signal described in detecting step a.

Using the method for signal detection in multi-antenna digital wireless communication system provided in an embodiment of the present invention, the complexity for calculating the initial value of Q can be reduced, and then reduce the use of the hardware resource of receiver, reduce the time delay of detection.

Detailed description of the invention

Fig. 1 is multi-antenna digital wireless communication system block diagram in the prior art；

Fig. 2 is the flow chart that the initial value of evaluated error covariance matrix of all transmitting signals is calculated in the embodiment of the present invention two；

Fig. 3 is the overhaul flow chart of the 1st transmitting signal in the embodiment of the present invention two；

Fig. 4 is the 2nd overhaul flow chart to m-th transmitting signal in the embodiment of the present invention two；

Fig. 5 is the flow chart of the On Square-Rooting Matrices of the inverse matrix of calculating matrix R in the embodiment of the present invention four；

Fig. 6 is the flow chart of the initial value for the evaluated error covariance matrix that the embodiment of the present invention five calculates all transmitting set of symbols；

Fig. 7 is the overhaul flow chart of the 1st transmitting signal in the embodiment of the present invention five；

Fig. 8 is the 2nd overhaul flow chart to m-th transmitting signal in the embodiment of the present invention five；

Fig. 9 is the flow chart that the estimated value of all transmitting signals is detected in the embodiment of the present invention six；

Fig. 1 O is the flow chart that the estimated value of all transmitting signals is detected in the embodiment of the present invention seven；

Figure 11 is the flow chart that the inverse matrix of general M × Metzler matrix is calculated in the embodiment of the present invention six；

Figure 12 is the flow chart of the LDU factoring matrix for the inverse matrix that general M × Metzler matrix is calculated in the embodiment of the present invention seven.

Specific embodiment

To make the objectives, technical solutions, and advantages of the present invention clearer, specific embodiment is named, the present invention is described in further detail.

The embodiment of the present invention one provides a kind of method that signal is detected in multi-antenna digital wireless communication system, the at least two transmitting signal of detection in multiple-input, multiple-output mimo system, the transmitting signal is emitted respectively by each different transmitting antenna of transmitting terminal and reaches receiving end by a channel, this method comprises:

A. at least two receiving antennas of receiving end receive the transmitting signal, obtain at least two and receive signal；

B. receiving end carries out channel estimation according to signal is received, and obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna；

C. the evaluated error covariance matrix of the partial transmitting signal in all transmitting signals is calculated using channel matrix H, then the evaluated error covariance matrix of calculated partial transmitting signal and the newly-increased matrix entries obtained according to the evaluated error covariance matrix and channel matrix H of the partial transmitting signal are utilized, recursion is acquired including the partial transmitting signal and number is more than the evaluated error covariance matrix of the transmitting signal of the partial transmitting signal number, the newly-increased matrix entries be include the newly-increased matrix entries of the partial transmitting signal and number more than the evaluated error covariance matrix of the evaluated error covariance matrix partial transmitting signal of the transmitting signal of the partial transmitting signal number；

D. the obtained evaluated error covariance matrix of step c is utilized, includes the transmitting signal that partial transmitting signal and number are more than the partial transmitting signal number described in detecting step c.

The method that second embodiment of the present invention provides a kind of to detect signal in multi-antenna digital wireless communication system.

Fig. 2 show the flow chart for calculating the initial value of evaluated error covariance matrix of all transmitting signals, including the following steps:

Step 201: receiving end receives the M transmitting signal that transmitting terminal emits respectively from M transmitting antenna, obtains N number of reception signal, and carry out channel estimation according to signal is received, obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna.

The sequencing that all M transmitting signals are detected in receiving end is preset, is denoted as t with transmitting signal serial number_M, t_M-1..., t_m..., t₂, t₁.Correspondingly, obtaining the channel matrix H of the sequencing sequence according to the pre-set detection transmitting signal channel matrix H by column rearrangement, being denoted as $H_M^{\left(t_M\right)}=[h_{{:t}_1}{h}_{{:\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}\&CenterDot;\&CenterDot;\&CenterDot;h_{:t_{M-1}}{h}_{{:t}_M}].$ Utilize vector f=[t₁, t₂..., t_m..., t_M-1, t_M]^TRecord and channel matrixThe index of corresponding transmitting signal.

Step 202: using channel matrix

First acquire

Cross-correlation channel matrix ${\mathrm{\&Phi;}}_M={\left(H_M^{\left(t_M\right)}\right)}^H\&CenterDot;H_M^{\left(t_M\right)},$ Again by Φ_MAcquire estimation error covariance matrix Q_MInverse matrix $R_M={\left(H_M^{\left(t_M\right)}\right)}^H\&CenterDot;H_M^{\left(t_M\right)}+\mathrm{\&alpha;}{I}_{M\&times;M}={\mathrm{\&Phi;}}_M+\mathrm{\&alpha;}{I}_{M\&times;M}.$

Wherein, $R_M={\left(H_M^{\left(t_M\right)}\right)}^H\&CenterDot;H_M^{\left(t_M\right)}+\mathrm{\&alpha;}{I}_{M\&times;M}$

Wherein, * expression takes conjugation to 1 plural number.

Step 203: calculating a finally detected transmitting signal t₁Corresponding evaluated error covariance matrix, is denoted as

Corresponding transmitting signal t₁Channel matrix be $H_1^{\left(t_1\right)}=\left[h_{{:t}_1}\right].$ The R obtained from step 202_MIn, obtain transmitting signal t₁The inverse matrix of evaluated error covariance matrix be $R_1^{\left(t_1\right)}={\left(h_{{:t}_1}\right)}^H\&CenterDot;h_{{:\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}+\mathrm{\&alpha;}=r_{t_1{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">t</figure-callout>}}_1},$ Wherein,

It is exactly R_MThe 1st row the 1st column element.

Then it utilizesCalculate corresponding finally detected transmitting signal t₁Estimation error covariance matrixBy $Q_1^{\left(t_1\right)}={\left(R_1^{\left(t_1\right)}\right)}^{-1}$ It obtains $Q_1^{\left(t_1\right)}=1/r_{t_1{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}.$

It allows the variable m of the initial value of recursion Q to be set as 2 below, enters step 204.In the initial value of following recursion Q the step of, m transmitting signal t that will be finally detected_m..., t₁, t₁Corresponding evaluated error covariance matrix is denoted as

Step 204: judging whether to have obtained the M-1 evaluated error covariance matrixes for being detected transmitting signal, that is, judge whether m is greater than M-1, if it is, having obtained the M-1 evaluated error covariance matrixes for being detected transmitting signal, go to step 208；Otherwise, recursion m are detected the evaluated error covariance matrix for emitting signal

Value, execute step 205,206,207.

Step 205: finally m detected transmitting signal t_m..., t₂, t₁Corresponding channel matrix is $H_m^{\left(t_m\right)}=[h_{{:t}_1}{h}_{{:\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}\&CenterDot;\&CenterDot;\&CenterDot;h_{{:t}_m}],$ Correspondingly, corresponding m finally detected transmitting signal t_m..., t₂, t₁The inverse matrix of evaluated error covariance matrix be $R_m^{\left(t_m\right)}={\left(H_m^{\left(t_m\right)}\right)}^H\&CenterDot;H_m^{\left(t_m\right)}+\mathrm{\&alpha;}{I}_{\left(m\right)\&times;\left(m\right)}.$

With

There is following recurrence relation:

$R_m^{\left(t_m\right)}=\left[\begin{array}{cc}{R}_{m-1}^{\left(t_{m-1}\right)}& v_{m-1}^{\left(t_m\right)}\\ {\left(v_{m-1}^{\left(t_m\right)}\right)}^H& {\mathrm{\&beta;}}_1^{\left(t_m\right)}\end{array}\right],$ Wherein,It is the m-1 transmitting signal t that correspondence that last recursion obtains finally is detected_m-1..., t₂, t₁Evaluated error covariance matrix the either corresponding 1 finally detected transmitting signal t of inverse matrix₁Estimation error covariance matrix inverse matrix

${\mathrm{\&beta;}}_1^{\left(t_m\right)}={h_{{:t}_m}}^H\&CenterDot;h_{{:t}_m}+\mathrm{\&alpha;}=r_{t_m{t}_m};$ $v_{m-1}^{\left(t_m\right)}=\left[\begin{array}{c}{h_{{:t}_1}}^H\&CenterDot;h_{{:t}_m}\\ {h_{{:t}_1}}^H\&CenterDot;h_{{:t}_m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {h_{{:t}_{m-1}}}^H\&CenterDot;h_{{:t}_m}\end{array}\right]=\left[\begin{array}{c}{r}_{t_1{t}_m}\\ r_{t_2{t}_m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ r_{t_{m-1}{t}_m}\end{array}\right].$

It can be seen that

With

The R that can be calculated from step 202_MIn directly obtain, more specifically,

It is R_MM row m column element, and

It is by R_MM column m-1, head form.Any calculating can thus not needed, so that it may directly obtain

Step 206: seeking m finally detected transmitting signal t_m..., t₂, t₁Corresponding evaluated error covariance matrix

Utilize the evaluated error covariance matrix of a finally detected transmitting signal of the correspondence acquired in step 203

Or the m-1 transmitting signal t that the obtained correspondence of the last recursion of the i.e. step 206 of this step is finally detected_m-1..., t₂, t₁Evaluated error covariance matrix

And it is obtained using step 205

WithRecursion obtainsRecurrence method is as described below:

First by

With

It calculates ${\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}=-Q_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}$ With ${\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}-{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{Q}_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}};$

Then by

With

Pass through $T_{m-1}^{\left(t_m\right)}=Q_{m-1}^{\left(t_{m-1}\right)}+{\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H$ It calculates

Thank to Germania-Morrison (Sherman-Morrison) result

Finally by $Q_m^{\left(t_m\right)}=\left[\begin{array}{cc}{T}_{m-1}^{\left(t_m\right)}& {\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\\ {\left({\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}\right)}^{*}{\left({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H& {\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}\end{array}\right]$ It obtains

From above method as can be seen that mathematically, we obtain

WithMeet such relationship:

$Q_m^{\left(t_m\right)}=\left[\begin{array}{cc}{Q}_{m-1}^{\left(t_{m-1}\right)}+{\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}{\mathrm{\&mu;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}\\ {\mathrm{\&chi;}}_{m-1}^{*}{\mathrm{\&mu;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}\end{array}\right],$ Pay attention in this expression formula, in some places, upper right mark (t is omitted in we_m)。

And when specific implementation, in order to reduce calculation amount, it is above-mentioned by

It obtains

Recurrence method can refine are as follows:

First by

With

It calculates ${\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}=-Q_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)},$ Again by

It calculates ${\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}+{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}};$

Then by

With

Pass through $T_{m-1}^{\left(t_m\right)}=Q_{m-1}^{\left(t_{m-1}\right)}+{\mathrm{\&omega;}}_{m-1}^{\left(t_{m-1}\right)}{\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H$ It calculates

Thank to Germania-Morrison (Sherman-Morrison) result

Finally by $Q_m^{\left(t_m\right)}=\left[\begin{array}{cc}{T}_{m-1}^{\left(t_m\right)}& {\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\\ {\left({\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}\right)}^{*}{\left({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H& {\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}\end{array}\right]$ It obtains

The value of step 207:m increases by 1, i.e. m=m+1, then return step 204.

Step 208: having obtained M-1 transmitting signal t now_M-1..., t_m..., t₂, t₁Corresponding evaluated error covariance matrixValue.First by

With

It calculates ${\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}=-Q_{M-1}^{\left(t_{M-1}\right)}{v}_{M-1}^{\left(t_M\right)}$ With ${\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_M\right)}-{\left(v_{M-1}^{\left(t_M\right)}\right)}^H{Q}_{M-1}^{\left(t_{M-1}\right)}{v}_{M-1}^{\left(t_M\right)}};$ When specific implementation, in order to reduce calculation amount, above-mentioned calculating process can be refined are as follows: first by

With

It calculates ${\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}=-Q_{M-1}^{\left(t_{M-1}\right)}{v}_{M-1}^{\left(t_M\right)},$ Again by

It calculates ${\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_M\right)}+{\left(v_{M-1}^{\left(t_M\right)}\right)}^H{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}}.$

Then, by

With

It calculatesDiagonal entry, i.e.,

$T_{M-1}^{\left(t_M\right)}(i,i)=Q_{M-1}^{\left(t_{M-1}\right)}(i,i)+{\mathrm{\&omega;}}_{M-1}^{\left(t_{M-1}\right)}{\mathrm{\&mu;}}_{M-1}^{\left(t_{M-1}\right)}\left(i\right){\left({\mathrm{\&mu;}}_{M-1}^{\left(t_{M-1}\right)}\left(i\right)\right)}^H,$

Wherein i=1,2 ..., M-1；

With

Respectively indicate matrix

With

I-th of element on diagonal line, i.e. matrix

With

Element on i-th row i-th column；And

Indicate column vector

I-th.

M=M is enabled in expression formula above-mentioned, we obtain expression formula $Q_M^{\left(t_M\right)}=\left[\begin{array}{cc}{T}_{M-1}^{\left(t_M\right)}& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\\ {\left({\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}\right)}^{*}{\mathrm{\&mu;}}_{M-1}^H& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}\end{array}\right],$ We do not need to calculate this expression formula, and only by this expression formula, we be can be clearly seen that, by the above-mentioned calculating process of this step, matrix is had been obtained in we

Diagonal line on all elements, and they are exactly

And matrix

Diagonal entry

I=1,2 ..., M-1.

Because the set of all M transmitting signals be it is identical, under the scene of ideal Floating-point Computation, all M evaluated error covariance matrixes for emitting signals are unique；And the sequencing t that pre-set all M transmitting signals are detected in receiving end_M, t_M-1..., t_m..., t₂, t₁Difference only results in eachCorresponding exchange is done in the position of interior row and column, and each

The element set for inside including is identical.So we can remember $Q_M^{\left(t_M\right)}=Q_M.$

The step of commencing signal detects below notices that we have acquired the evaluated error covariance matrix of M all transmitting signals $Q_M=Q_M^{\left(t_M\right)}$ Diagonal line on all elements, and

It is exactly that optimal detection sequence is determined in signal detection process, and when gradually detecting according to the optimal detection sequence and using interference elimination method each transmitting signal, the initial value of used evaluated error covariance matrix.Furthermore we acquire during above-mentioned recursion

With it is each

M=1,2 ..., M-1, described is eachIt is the evaluated error covariance matrix of part transmitting signal to be detected, is corresponding transmitting antenna t_m..., t₂, t₁Evaluated error covariance matrix.

After obtaining the initial value of the evaluated error covariance matrix of all transmitting signals to be detected, into the process of detection signal shown in Fig. 3, that is, a of Fig. 3 is gone to.

Fig. 3 is the flow chart of signal detection, and signal detection shown in Fig. 3 is since a.

Step 301: carrying out pre-matching filtering transformation to signal is received, obtain the pre-matching filter result for receiving signal $z_M={\left(H_M\right)}^H\&CenterDot;\stackrel{\&RightArrow;}{x},$ Wherein, (H_M)^HFor matched filter, vectorTo indicate to receive signal x₁..., x_NVector.The index for emitting signal is still vector f=[t₁, t₂..., t_m..., t_M-1, t_M]^T。

The iteration variable m that will test in signal process below is set as 1, enters step 302.

Step 302: emitting in signal at M and determine the best transmitting signal of received signal to noise ratio, method is: in Q_MDiagonal entry in, search the taken the smallest item of value, be denoted as l_MCapable and l_MThe item of column, i.e., $l_M=\underset{i=1}{\overset{M-m+1}{\arg \min }}{Q}_{M-m+1}(i,i).$ The Q_M+1-mL in diagonal entry_MCapable and l_MThe item of column corresponds to the signal that received signal to noise ratio is best in M+1-m transmitting signal, i.e. currently detected transmitting signal.

Step 303: calculated for subsequent signal detection needs the matrix Q used_MOff-diagonal element, they are exactly Q_ML_MCapable off diagonal element and l_MThe off diagonal element of column.According to matrix Q_MMeet $Q_M^H=Q_M$ Symmetric relation to Q_M(i, j)=(Q_M(j, i))^*, it is only necessary to calculate Q_ML_MThe off-diagonal element Q of column_M(i, l_M), i=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., M, and Q_ML_MCapable each off-diagonal element Q_M(l_M, i) and (i=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., M) it can be by Q_ML_MThe element Q of the corresponding position of column_M(i, l_M) conjugation is taken to obtain.

And Q_ML_MThe off-diagonal element of column, by above-mentioned

$T_{M-1}^{\left(t_M\right)}=Q_{M-1}^{\left(t_{M-1}\right)}+{\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}{\left({\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\right)}^H$ With $Q_M^{\left(t_M\right)}=\left[\begin{array}{cc}{T}_{M-1}^{\left(t_M\right)}& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\\ {\left({\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}\right)}^{*}{\left({\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\right)}^H& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}\end{array}\right]$ It calculates.Specifically, if l_M=M, then Q_MM column off diagonal element by

It provides；And if l_M≠ M, then Q_ML_MThe off diagonal element of column respectively by

$Q_M^{\left(t_M\right)}(i,l_M)=T_{M-1}^{\left(t_M\right)}(i,l_M)=Q_{M-1}^{\left(t_{M-1}\right)}(i,l_M)+{\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\left(i\right){\left({\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\left(l_M\right)\right)}^{*},$

I=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., M-1,

With $Q_M^{\left(t_M\right)}(M,l_M)={\left({\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}\right)}^{*}{\left({\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\left(l_M\right)\right)}^{*}$ It calculates.

Step 304: in matrix Q_MMiddle exchange l_MCapable and M row exchanges l_MColumn and m column；Correspondingly, in matrix Φ_MMiddle exchange l_MCapable and M row exchanges l_MColumn and m column；Correspondingly, to the pre-matching filter result vector z for receiving signal_MMiddle exchange l_MAnd M.L is exchanged in vector f_MAnd M.Pay attention to matrix Q_MIn M row and the off-diagonal element of m column all do not found out to come in most cases, but this is not interfered in Q_MThe middle exchange for carrying out above-mentioned row and column.

Below with Q_MFor column, illustrate to carry out the result after the exchange of above-mentioned row and column.

In matrix Q_MMiddle exchange l_MIt is obtained after capable and M row

In matrix Q_MMiddle exchange l_MIt is obtained after column and m column

It can easily be seen that the Q obtained after the exchange of above-mentioned row and column "_MM column be $q_M=\left[\begin{array}{c}{Q}_M(1,l_M)\\ Q_M(2,l_M)\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ Q_M(l_M-1,l_M)\\ Q_M(M,l_M)\\ Q_M(l_M+1,l_M)\\ Q_M(l_M+2,l_M)\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ Q_M(M-1,l_M)\\ Q_M(l_M,l_M)\end{array}\right],$ q_MIt will be used for subsequent signal detection, and be readily seen that q_MElement be all row and column exchange before Q_MIn l_MThe element of column, to find out in step 303.

Step 305: currently detected transmitting signal is to be denoted as p M+1-m in vector f_m, p_m=f (M+1-m).Calculate the estimated value to currently detected transmitting signalEmit signal p_mEstimated value

For, $y_{p_m}=q_M^H\&CenterDot;z_M.$ Wherein, q_MIndicate Q_MM column, as previously described.

Step 306: to the estimated value of obtained transmitting signal

Quantified, obtains the testing result to transmitting signal

Step 307: the influence of currently detected transmitting signal is eliminated from the pre-matching filter result for receiving signal, the corresponding multiple pre-matching filter results for receiving signal of M-m transmitting signal being not yet detected, it may be assumed that the pre-matching filter result vector z for receiving signal_M+1-mMiddle deletion M+1-m obtains the (z with M-m_M+1-m)^minus；From (z_M+1-m)^minusThe middle interference for eliminating the transmitting signal being currently detected obtains the pre-matching filter result z for corresponding to multiple reception signals of all M-m not detected transmitting signals_M-m, $z_{M-m}={\left(z_{M+1-m}\right)}^{\min \mathrm{us}}-{\hat{s}}_{p_m}\&CenterDot;{\mathrm{\&phi;}}_{M-\mathrm{<figure-callout\; id="1"\; label="m\; +"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">m</figure-callout>}+1},$ Wherein φ_M-m+1It is matrix Φ_M-m+1M+1-m column head M-m row, φ_M-m+1With M-m.

Step 308: from matrix Φ_M+1-mMiddle deletion M+1-m column and M+1-m row, obtain matrix Φ_M-m。

Start the 2nd below, 3 ... until the detection of m-th transmitting signal goes to the b of Fig. 4 into the process of detection signal shown in Fig. 4, it is constant to pay attention to that the iteration variable m in detection signal process is still set as 1.

Step 401: judging to seek Q in the embodiment of the present invention two_MInitial value during, if obtained the corresponding M-m evaluated error covariance matrix Q for emitting signal_M-m, i.e. judgment step 203 is obtained

It is obtained each with being performed a plurality of times for step 206

In (m=2,3 ..., M-1), if havingSet [the t of M-m corresponding transmitting signal₁ t₂ … t_M-m-1 t_M-m] with current detection signal needed for Q_M-mThe set of corresponding M-m transmitting signal is identical set not considering under the premise of the putting in order of each element in set；If so, thening follow the steps 402；Otherwise, step 403 is executed.

As previously mentioned, in the prior art $Q_{M+1-m}=\left[\begin{array}{cc}{T}_{M-m}& w_{M-m}\\ {\left(w_{M-m}\right)}^H& {\mathrm{\&omega;}}_{M-m}\end{array}\right],$ And by $Q_{M-n}=T_{M-m}-\frac{1}{{\mathrm{\&omega;}}_{M-m}}{w}_{M-m}\&CenterDot;{\left(w_{M-m}\right)}^H$ Calculate Q_M-m.And the present invention only utilizes matrix Q_M+1-mCalculate the evaluated error covariance matrix Q that M-m transmitting signal is corresponded to required for next iteration_M-mIn required item, specific method includes step 403,404,405,406.

Step 402: by recursion Q_MInitial value during obtained Q_M-mAs the evaluated error covariance matrix Q for corresponding to M-m transmitting signal required for next iteration_M-m.Then step 407 is executed.

Step 403: first by T_M-mAnd w_M-mCalculate Q_M-mDiagonal entry, i.e.,

$Q_{M-m}(i,i)=T_{M-m}(i,i)-\frac{1}{{\mathrm{\&omega;}}_{M-m}}{w}_{M-m}\left(i\right)\&CenterDot;{\left(w_{M-m}\left(i\right)\right)}^{*},$ Wherein i=1,2 ..., M-m, Q_M-m(i, i) and T_M-m(i, i) respectively indicates matrix Q_M-mAnd T_M-mI-th of element on diagonal line, i.e. matrix Q_M-mAnd T_M-mElement on i-th row i-th column, and w_M-m(i) column vector w is indicated_M-mI-th.

Step 404: determining the best transmitting signal of received signal to noise ratio in M-m transmitting signals to be detected, method is: in Q_M-mDiagonal entry in, search the taken the smallest item of value, be denoted as l_M-mCapable and l_M-mThe item of column, i.e., $l_{M-m}=\underset{i=1}{\overset{M-m}{\arg \min }}{Q}_{M-m}(i,i).$ The Q_M-mL in diagonal entry_M-mCapable and l_M-mThe item of column corresponds to the signal that received signal to noise ratio is best in M-m transmitting signal, i.e. currently detected transmitting signal.

Step 405: calculated for subsequent signal detection needs the matrix Q used_M-mOff-diagonal element, they are exactly Q_M-mL_M-mCapable off diagonal element and l_M-mThe off diagonal element of column.According to matrix Q_M-mMeet $Q_{M-m}^H=Q_{M-m}$ Symmetric relation to Q_M-m(i, j)=(Q_M-m(j, i))^*, it is only necessary to calculate Q_M-mL_M-mThe off-diagonal element Q of column_M-m(i, l_M-m), i=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., M-m, and Q_M-mL_M-mCapable each off diagonal element Q_M-m(l_M-m, i) and (i=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., M-m) it can be by Q_M-mL_M-mThe element Q of the corresponding position of column_M-m(i, l_M-m) conjugation is taken to obtain.

And Q_M-mL_M-mThe off diagonal element of column, by

$Q_{M-m}(i,l_{M-m})=T_{M-m}(i,l_{M-m})-\frac{1}{{\mathrm{\&omega;}}_{M-m}}{w}_{M-m}\left(i\right)\&CenterDot;{\left(w_{M-m}\left(l_{M-m}\right)\right)}^{*}$ It calculates, wherein i=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., M-m.And T_M-m(i, l_M-m) it is matrix Q_M-m+1In one, if Q_M-m+1It is known that then directly from Q_M-m+1Obtain T_M-m(i, l_M-m)；Otherwise Q is acquired with following methods_M-m+1In this value: successively look for Q_M-m+2、Q_M-m+3、...、Q_MIn respective items whether it is known that until in first Q_M-m+k, it is known for finding a respective items in k=2 ..., m, then thus according to shown in this step by Q_i+1Seek Q_iMethod, by Q_M-m+kOne respective items, successively acquire matrix Q_M-m+k-k1, respective items in k1=1,2 ..., k-1 finally acquire Q_M-m+1In respective items be exactly required T_M-m(i, l_M-m)；If until Q_MIn respective items be not still it is known, then the method shown in step 303 acquires Q_MIn respective items value, then thus according to shown in this step by Q_i+1Seek Q_iMethod, by Q_MOne respective items, successively acquire matrix Q_M-k2, respective items in k2=1,2 ..., m-1 finally acquire Q_M-m+1In respective items be exactly required T_M-m(i, l_M-m)。

Step 406: in matrix Q_M-mMiddle exchange l_M-mCapable and M-m row exchanges l_M-mColumn and M-m column；Correspondingly, in matrix Φ_M-mMiddle exchange l_M-mCapable and M-m row exchanges l_M-mColumn and M-m column；Correspondingly, to the pre-matching filter result vector z for receiving signal_M-mMiddle exchange l_M-mAnd M-m.L is exchanged in vector f_M-mAnd M-m.

Similar to the principle in abovementioned steps 304, matrix Q_M-mIn the off-diagonal element of M-m row and M-m column all do not found out to come in most cases, but this is not interfered in Q_M-mThe middle exchange for carrying out above-mentioned row and column.And Q_M-mAfter the exchange of above-mentioned row and column, last column are that M-m is classified as $q_{M-m}=\left[\begin{array}{c}{Q}_{M-m}(1,l_M)\\ Q_{M-m}(2,l_{M-m})\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ Q_{M-m}(l_{M-m}-1,l_{M-m})\\ Q_{M-m}(M-m,l_{M-m})\\ Q_{M-m}(l_{M-\mathrm{<figure-callout\; id="1"\; label="m\; +"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">m</figure-callout>}}+1,l_{M-m})\\ Q_{M-m}(l_{M-\mathrm{<figure-callout\; id="2"\; label="m\; +"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">m</figure-callout>}}+2,l_{M-m})\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ Q_{M-m}(M-m-1,l_{M-m})\\ Q_{M-m}(l_{M-m},l_{M-m})\end{array}\right],$ It can see q_M-mElement be all row and column exchange before Q_M-mIn l_M-mThe element of column, thus found out in step 405, and q_M-mFor subsequent signal detection.

Step 407: currently detected transmitting signal is to be denoted as p M-m in vector f_m+1, p_m+1=f (M-m).Calculate the estimated value to currently detected transmitting signal

Emit signal p_m+1Estimated value

For, $y_{p_{m+1}}=q_{M-m}^H\&CenterDot;z_{M-m}.$ Wherein, q_M-mIndicate Q_M-mM-m column.

Step 408: to the estimated value of obtained transmitting signal

Quantified, obtains the testing result to transmitting signal

Step 409: the influence of currently detected transmitting signal is eliminated from the pre-matching filter result for receiving signal, the corresponding multiple pre-matching filter results for receiving signal of M-m-1 transmitting signal being not yet detected, it may be assumed that the pre-matching filter result vector z for receiving signal_M-mMiddle deletion M-m obtains the (z with M-m-1_M-m)^min us；From (z_M-m)^min usThe middle interference for eliminating the transmitting signal being currently detected obtains the pre-matching filter result z for corresponding to multiple reception signals of all M-m-1 not detected transmitting signals_M-(m+1), $z_{M-m-1}={\left(z_{M-m}\right)}^{\min \mathrm{us}}-{\hat{s}}_{p_{m+1}}\&CenterDot;{\mathrm{\&phi;}}_{M-m},$ Wherein φ_M-mIt is matrix Φ_M-mM-m column head M-m-1 row, φ_M-mWith M-m-1.

Step 410: from matrix Φ_M-mMiddle deletion M-m column and M-m row, obtain matrix Φ_M-m-1。

Step 411: judging whether be in next step the last one transmitting signal of detection, that is, judge whether m is equal to M-2, if so, thening follow the steps 413；It is no to then follow the steps 412.

The value increase by 1 of step 412:m, i.e. m=m+1, return step 401.

Step 413: currently detected transmitting signal is the first item of vector f, is denoted as p_M, p_M=f (1).Calculate the estimated value of finally detected transmitting signal

Emit signal p_MEstimated value

For, $y_{p_M}=q_1^H\&CenterDot;{\mathrm{<figure-callout\; id="1"\; label="z"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">z</figure-callout>}}_1.$ Wherein, q₁It is exactly Q₁。

Step 414: to the estimated value of obtained transmitting signal

Quantified, obtains the testing result to transmitting signal

By above method, the sequencing detected to transmitting signal is p₁, p₂..., p_M, corresponding, the testing result for emitting signal is

It can see in embodiment described above, if recursion emit signal evaluated error covariance matrix Q initial value when institute it is pre-set detection emit signal sequencing it is identical as the actually detected transmitting sequence of signal, some elements in evaluated error covariance matrix needed for next iteration is calculated in the method as described in step 403 to 406 are not had to then, and evaluated error covariance matrix needed for directly obtained intermediate result obtains next iteration during the initial value of recursion Q, therefore, calculation amount can be much less；Alternatively, if when the initial value of recursion Q pre-set detection transmitting signal sequencing it is close with the actually detected transmitting sequence of signal, the possibility for omitting the step 403 to 406 is also bigger, thus, it is also possible to reduce many calculation amounts.

To, in slow fading channel, the sequencing that M transmitting antenna is detected in receiving end can be set to the optimal ordering of the last detection, so that in detection signal process shown in Fig. 4, increase the probability that the judging result of the step 401 is yes, accordingly reduces step 403 to the 406 bring calculation amount.Because in slow fading channel, changes in channel characteristics is slow, compared with the last optimal detection sequence, the optimal detection sequence at current time changes calculation amount that is little or identical, therefore can be good at obtained intermediate result reduction detection signal during the initial value using recursion Q.

In fast fading channel, a detection ordering can also be estimated by channel matrix H, so that optimal detection sequence of this detection ordering close to actual use.

In some applications, the sequences of all transmitting signals of detection are fixed in advance, and the detection ordering fixed in advance according to this detects transmitting signal one by one, do not need to ask optimal detection ordering in the process.In this case, the sequencing that all M transmitting signals pre-set in the step 201 are detected in receiving end, it is exactly the detection ordering fixed in advance, in this way, the process of the determination detection ordering as described in step 303 is not needed, but currently selects which signal to be detected to be detected according to the detection ordering determination fixed in advance；Accordingly, it is not necessary to the exchange of the row and column of the matrix as described in step 304；Meanwhile the process of evaluated error covariance matrix needed for not needing the calculating next iteration as described in step 403 to 406, directly next transmitting signal to be detected is detected using intermediate result obtained during recursion Q.By above method, calculation amount can be much less.

Therefore, the embodiment of the present invention can be by first finding out Q_M-mWhich element needs used during signal detection, then these elements are calculated, to avoid Q is calculated_M-mThe unnecessary calculation amount that generates of the element that is not used in signal detection process.

The embodiment of the present invention three additionally provides a kind of method that signal is detected in multi-antenna digital wireless communication system.

On the basis of the embodiment of the present invention two, the step 206 of the embodiment of the present invention two is changed to following step 206 ': seek m finally detected transmitting signal t_m..., t₂, t₁Corresponding evaluated error covariance matrix

Utilize the evaluated error covariance matrix of a corresponding finally detected transmitting signal

Or the m-1 transmitting signal t that the obtained correspondence of last recursion is finally detected_m-1..., t₂, t₁Evaluated error covariance matrix

Recursion obtains

Recurrence method is as described below:

First by

With

It utilizes ${\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}\right)}^{*}={\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}-{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{Q}_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}}$ Relationship calculate one

Such as calculate one ${\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}=\sqrt{\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}-{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{Q}_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}}},$ Simultaneously also by ${\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}={\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\left(t_{m-1}\right)}=-{\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}{Q}_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}$ It calculates

Then by

Pass through $T_{m-1}^{\left(t_m\right)}=Q_{m-1}^{\left(t_{m-1}\right)}+{\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}{\left({\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}\right)}^H$ It calculates

Thank to Germania-Morrison (Sherman-Morrison) result

Thus by $Q_m^{\left(t_m\right)}=\left[\begin{array}{cc}{T}_{m-1}^{\left(t_m\right)}& {\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}\\ {\left({\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}\right)}^H& {\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}\right)}^{*}\end{array}\right]$ It obtains

From above method as can be seen that mathematically, we obtain

With

Meet such relationship: $Q_m^{\left(t_m\right)}=\left[\begin{array}{cc}{Q}_{m-1}^{\left(t_{m-1}\right)}+{\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}{\left({\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}\right)}^H& {\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}\\ {\left({\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}\right)}^H& {\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}\right)}^{*}\end{array}\right].$

And when specific implementation, in order to reduce calculation amount, it is above-mentioned by

It obtains

Recurrence method can refine are as follows:

First by

With

It calculates ${\mathrm{\&mu;}}_{m-1}^{\left(t_{m-1}\right)}=-Q_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)},$ Again by

It calculates ${\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}\right)}^{*}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}+{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{\mathrm{\&mu;}}_{m-1}^{\left(t_{m-1}\right)}}$ And ${\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}=\sqrt{{\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}\right)}^{*}},$ Then by ${\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}={\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\left(t_{m-1}\right)}$ It calculates

Finally by

Pass through $T_{m-1}^{\left(t_m\right)}=Q_{m-1}^{\left(t_{m-1}\right)}+{\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}{\left({\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}\right)}^H$ It calculates

Thank to Germania-Morrison (Sherman-Morrison) result

Thus by $Q_m^{\left(t_m\right)}=\left[\begin{array}{cc}{T}_{m-1}^{\left(t_m\right)}& {\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}\\ {\left({\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}^{\left(t_{m-1}\right)}\right)}^H& {\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&chi;}}_{m-1}^{\left(t_m\right)}\right)}^{*}\end{array}\right].$ It obtains

More generally, the embodiment of the present invention four has also provided the method for seeking an inverse of a matrix matrix, and computation complexity needed for this method is lower than existing method.

In the embodiment of the present invention, the method for calculating the On Square-Rooting Matrices of M × Metzler matrix inverse matrix is provided.M × the Metzler matrix is expressed as

Matrix R meets: the associate matrix R of matrix R and the matrix^HIt is equal.

The step of On Square-Rooting Matrices of the inverse matrix of calculating matrix R are given below, realizes especially by following steps, as shown in Figure 5:

Step 501: calculating R⁽¹⁾=[r₁₁] inverse matrix (R⁽¹⁾)^-1, i.e. Q⁽¹⁾=(R⁽¹⁾)^-1。

Wherein, R⁽¹⁾Be R matrix preceding 1 row before 1 column element constitute submatrix.In the present embodiment, the initial submatrix for recursion is the submatrix of 1 row 1 column.Certainly, in practical applications, the submatrix of (k < M) can be arranged as the initial submatrix for being used for recursion down to arbitrary k row k using 2 rows 2 column.

Recursion variable m=2 is set.

Step 502: judging whether to have obtained the factoring matrix of R inverse of a matrix matrix, that is, judge whether m is greater than M, if it is, going to step 506；Otherwise, step 503 is gone to.

Step 503: obtaining the R of this recursion^(m)Matrix, R^(m)It can be in the R of upper primary recursion^(m-1)On the basis of increase a line, one column obtain, be embodied as $R^{\left(m\right)}=\left[\begin{array}{cc}{R}^{(m-1)}& v_{m-1}\\ {\left(v_{m-1}\right)}^H& {\mathrm{\&beta;}}_m\end{array}\right].$

Wherein, R^(m)Matrix is the submatrix of the preceding m row m column of R matrix, specially

R^(m-1)Matrix is the submatrix of the preceding m-1 row m-1 column of R matrix, specially

v_m-1For R matrix m arrange preceding m-1 form vectors, specially $v_{m-1}=\left[\begin{array}{c}{r}_{1m}\\ {\mathrm{<figure-callout\; id="2m"\; label="r"\; filenames="H2008101874159E00091.png"\; state="\{\{state\}\}">r</figure-callout>}}_{2m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ r_{(m-1)m}\end{array}\right];$ β_mFor the element that the m row m of R matrix is arranged, specially r_mm。

Submatrix R for recursion^(m)It needs to extract from the diagonal line of matrix R, i.e. the R of the submatrix^(m)Meet: submatrix R^(m)With the associate matrix (R of the submatrix^(m))^HIt is equal.

In actual matrix decomposable process, do not need specifically to obtain R^(m)Complete structure, and only need to obtain v_m-1And β_m, v_m-1And β_mIt can directly extract and obtain from matrix R.

Step 504: calculating R^(m)Inverse matrix (R^(m))^-1=Q^(m)。

First by Q_m-1、v_m-1And β_mCalculate μ_m-1=-Q_m-1v_m-1With ${\mathrm{\&omega;}}_{m-1}=\frac{1}{{\mathrm{\&beta;}}_m-{\left(v_{m-1}\right)}^H{Q}_{m-1}{v}_{m-1}},$ Again by μ_m-1And ω_m-1Pass through T_m-1=Q_m-1+ω_m-1μ_m-1(μ_m-1)^HCalculate Q_m-1Thank to Germania-Morrison (Sherman-Morrison) result T_m-1, then by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}\\ {\mathrm{\&omega;}}_{m-1}^{*}{\mathrm{\&mu;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}\end{array}\right]$ Obtain Q_m。

And when specific implementation, it is above-mentioned by Q in order to reduce calculation amount_m-1Obtain Q_mRecurrence method can refine are as follows:

First by Q_m-1、v_m-1And β_mCalculate μ_m-1=-Q_m-1v_m-1, then by μ_m-1It calculates ${\mathrm{\&omega;}}_{m-1}=\frac{1}{{\mathrm{\&beta;}}_m+{\left(v_{m-1}\right)}^H{\mathrm{\&mu;}}_{m-1}},$ Finally by μ_m-1And ω_m-1Pass through T_m-1=Q_m-1+ω_m-1μ_m-1(μ_m-1)^HCalculate Q_m-1Thank to Germania-Morrison (Sherman-Morrison) result T_m-1, then by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}\\ {\mathrm{\&omega;}}_{m-1}^{*}{\mathrm{\&mu;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}\end{array}\right]$ Obtain Q_m。

The value of step 505:m increases by 1, i.e. m=m+1 then goes to step 502.

In recursive process, recursion variable m can also increase integer, such as m=m+2 or m=m+3 greater than 1 etc..It means that if the inverse matrix for calculating 2 × 2 submatrixs can directly calculate the inverse matrix of 4 × 4 submatrixs, or can directly calculate the inverse matrix of 5 × 5 submatrixs then in recursive process next time in this recursive process.

Step 506: obtaining the inverse matrix R of matrix R^-1=Q, obtained Q meet Q=R^-1Relationship.

The embodiment of the present invention four additionally provides a kind of method for seeking an inverse of a matrix matrix.

The step 504 of the embodiment of the present invention is revised as steps described below 504 ', and other steps of embodiment three all remain unchanged.Modify obtain step 504 ' it is as follows.

Step 504 ': calculate R^(m)Inverse matrix (R^(m))^-1=Q^(m)。

First by Q_m-1、v_m-1And β_m, utilize ${\mathrm{\&chi;}}_{m-1}{\left({\mathrm{\&chi;}}_{m-1}\right)}^{*}={\mathrm{\&omega;}}_{m-1}=\frac{1}{{\mathrm{\&beta;}}_m-{\left(v_{m-1}\right)}^H{Q}_{m-1}{v}_{m-1}}$ Relationship calculate an x_m-1, such as calculate one ${\mathrm{\&chi;}}_{m-1}=\sqrt{\frac{1}{{\mathrm{\&beta;}}_m-{\left(v_{m-1}\right)}^H{Q}_{m-1}{v}_{m-1}}},$ Simultaneously also by μ_m-1=x_m-1μ_m-1=-x_m-1Q_m-1v_m-1Calculate μ_m-1；Then by μ_m-1Pass through T_m-1=Q_m-1+μ_m-1(μ_m-1)^HCalculate Q_m-1Thank to Germania-Morrison (Sherman-Morrison) result T_m-1, thus by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}\\ {\left({\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}\right)}^H& {\mathrm{\&chi;}}_{m-1}{\left({\mathrm{\&chi;}}_{m-1}\right)}^{*}\end{array}\right]$ Obtain Q_m。

And when specific implementation, it is above-mentioned by Q in order to reduce calculation amount_m-1Obtain Q_mRecurrence method can refine are as follows:

First by Q_m-1、v_m-1And β_mCalculate μ_m-1=-Q_m-1v_m-1, then by μ_m-1It calculates ${\mathrm{\&chi;}}_{m-1}{\left({\mathrm{\&chi;}}_{m-1}\right)}^{*}=\frac{1}{{\mathrm{\&beta;}}_m+{\left(v_{m-1}\right)}^H{\mathrm{\&mu;}}_{m-1}}$ And one ${\mathrm{\&chi;}}_{m-1}=\sqrt{{\mathrm{\&chi;}}_{m-1}{\left({\mathrm{\&chi;}}_{m-1}\right)}^{*}},$ Then by μ_m-1=x_m-1μ_m-1Calculate μ_m-1, finally by μ_m-1Pass through T_m-1=Q_m-1+μ_m-1(μ_m-1)^HCalculate Q_m-1Thank to Germania-Morrison (Sherman-Morrison) result T_m-1, thus by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}\\ {\left({\stackrel{\&OverBar;}{\mathrm{\&mu;}}}_{m-1}\right)}^H& {\mathrm{\&chi;}}_{m-1}{\left({\mathrm{\&chi;}}_{m-1}\right)}^{*}\end{array}\right]$ Obtain Q_m。

Method this method that the embodiment of the present invention five additionally provides a kind of detection signal in multi-antenna digital wireless communication system can be applied to the receiver in the case that the signal that transmitting terminal is emitted includes at least one set Alamouti Space-Time Block Coding.

Assuming that transmitting terminal has 2M transmitting antenna, every 2 transmitting antennas emit one group of Alamouti Space-Time Block Coding, then transmitting terminal emits M group Alamouti Space-Time Block Coding altogether, and wherein Alamouti Space-Time Block Coding is the channel signal that a group code is obtained by Alamouti Space-Time Block Coding encoder channel coding.There is N number of receiving antenna in receiving end, and the number of N is more than or equal to M.

In the communication system with the M group Alamouti Space-Time Block Coding, the channel matrix between the two emitting antennas and N number of receiving antenna of first group of transmission Alamouti Space-Time Block Coding is expressed as $\left(\begin{array}{cc}{h}_{11}^1& h_{12}^1\\ h_{21}^1& h_{22}^1\\ \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;\\ {\mathrm{<figure-callout\; id="1"\; label="h\; N"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">h</figure-callout>}}_N^1& {\mathrm{<figure-callout\; id="2"\; label="h\; N"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">h</figure-callout>}}_N^2\end{array}\right),$ Channel matrix between the two emitting antennas and N number of receiving antenna of second group of transmission Alamouti Space-Time Block Coding is expressed as $\left(\begin{array}{cc}{h}_{11}^2& h_{12}^2\\ h_{21}^2& h_{22}^2\\ \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;\\ {\mathrm{<figure-callout\; id="1"\; label="h\; N"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">h</figure-callout>}}_N^1& {\mathrm{<figure-callout\; id="2"\; label="h\; N"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">h</figure-callout>}}_N^2\end{array}\right).$ In general, channel matrix m group sent between the two emitting antennas and N number of receiving antenna of Alamouti Space-Time Block Coding is expressed as $\left(\begin{array}{cc}{h}_{11}^m& h_{12}^m\\ h_{21}^m& h_{22}^m\\ \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;\\ h_{N1}^m& {\mathrm{<figure-callout\; id="2m"\; label="h\; N"\; filenames="H2008101874159E00091.png"\; state="\{\{state\}\}">h</figure-callout>}}_N^2\end{array}\right),$ And the symbol that m group transmitting antenna emits is expressed as { s₁ ^m, s₂ ^m, wherein m=1,2 ..., M.

At this moment, in two symbol periods in an Alamouti Space-Time Block Coding period, the reception signal r on N number of receiving antenna of receiving end is r=Hs+ η, can be expressed as form:

Wherein, equivalent channel matrix H is the matrix of a 2N × 2M.

In addition, wherein there is L group Alamouti Space-Time Block Coding by 2L transmitting antenna transmitting, every 2 transmitting antennas emit one group of Alamouti Space-Time Block Coding, and the symbol emitted is expressed as s if transmitting terminal shares 2L+K transmitting antenna¹, s²... s^L, s is expressed as in the symbol that first symbol period in Alamouti Space-Time Block Coding period emits₁ ¹, s₁ ²... s₁ ^L, s is expressed as in the symbol that second symbol period in Alamouti Space-Time Block Coding period emits₂ ¹, s₂ ²... s₂ ^L；Meanwhile there are also K group codes directly to pass through K transmitting antenna transmitting, for every group code by a transmitting antenna transmitting, the symbol emitted is expressed as b¹, b²... b^K, b is expressed as in the symbol that first symbol period in Alamouti Space-Time Block Coding period emits₁ ¹, b₁ ²... b₁ ^K, b is expressed as in the symbol that second symbol period in Alamouti Space-Time Block Coding period emits₂ ¹, b₂ ²... b₂ ^K.There is N number of transmitting antenna in receiving end, meets N >=L+K.Wherein, the channel matrix k-th emitted between the single transmitting antenna and N number of receiving antenna of a group code is expressed as [f_1k f_2k … f_Nk], wherein k=1,2 ..., K.

At this moment, in two symbol periods in an Alamouti Space-Time Block Coding period, the reception signal r on N number of receiving antenna of receiving end is $r=\left[\begin{array}{cc}H1& F\end{array}\right]\left[\begin{array}{c}s1\\ b\end{array}\right]+\mathrm{\&eta;}=\mathrm{Hs}+\mathrm{\&eta;},$ Wherein,

$r=\left[\begin{array}{c}{r}_{11}\\ r_{12}^{*}\\ r_{21}\\ r_{22}^{*}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ r_{N1}\\ {\mathrm{<figure-callout\; id="2"\; label="r\; N"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">r</figure-callout>}}_N^2\end{array}\right],$

$\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">s</figure-callout>}1=\left[\begin{array}{c}{\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">s</figure-callout>}}_1^1\\ {\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">s</figure-callout>}}_2^1\\ {\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">s</figure-callout>}}_1^2\\ {\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">s</figure-callout>}}_2^2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ s_1^L\\ s_2^L\end{array}\right],$ $\mathrm{\&eta;}=\left[\begin{array}{c}{\mathrm{\&eta;}}_{11}\\ {\mathrm{\&eta;}}_{12}^{*}\\ {\mathrm{\&eta;}}_{21}\\ {\mathrm{\&eta;}}_{22}^{*}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {\mathrm{\&eta;}}_{\mathrm{<figure-callout\; id="1"\; label="N"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">N</figure-callout>}1}\\ {\mathrm{\&eta;}}_{\mathrm{<figure-callout\; id="2"\; label="N"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">N</figure-callout>}2}^{*}\end{array}\right];$

$b=\left[\begin{array}{c}{\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^1\\ {\left({\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^1\right)}^{*}\\ {\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^2\\ {\left({\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^2\right)}^{*}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ b_1^K\\ {\left(b_2^K\right)}^{*}\end{array}\right].$ F can also be rewritten into the matrix for having same format with H1, corresponding b is also required to rewrite, obtain F and b as described below:

$\left[\begin{array}{c}{\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^1\\ -{\left({\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^1\right)}^{*}\\ {\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^2\\ -{\left({\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^2\right)}^{*}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ b_1^K\\ -{\left(b_2^K\right)}^{*}\end{array}\right].$

The F and b of above two form, select it is any receptivity is not influenced, the also very little of the influence for Receiver Complexity, thus can ignore.The F and b of above two form, can also be following form:

$b=\left[\begin{array}{c}{\left({\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^1\right)}^{*}\\ {\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^1\\ {\left({\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^2\right)}^{*}\\ {\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {\left(b_1^K\right)}^{*}\\ b_2^K\end{array}\right];$

$b=\left[\begin{array}{c}{\left({\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^1\right)}^{*}\\ -{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^1\\ {\left({\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^2\right)}^{*}\\ {-\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {\left(b_1^K\right)}^{*}\\ -b_2^K\end{array}\right].$

The essence of the F and b of above-mentioned 4 kinds of forms are two symbols that the same transmitting antenna k emits respectively in two symbol periods in an Alamouti Space-Time Block Coding period, they correspond to two of the same receiving antenna n in F, necessarily (f_nk)^*With f_nkOr-(f_nk)^*With f_nkRelationship, i.e., be conjugated each other or negative conjugate relation.The performance of receiver can be improved in this building method.

In the embodiment of the present invention, by taking second of representation method of F as an example, signal detecting result is provided.At this moment, outside L group Alamouti Space-Time Block Coding plus in the channel model of K group code, channel matrix H is the matrix of 2N × 2 (L+K), as follows:

If by 2L+K all transmitting antennas, it is indicated with transmitting set of symbols, wherein 2L transmitting antenna set of symbols 1,2, ..., L expression, K transmitting antenna set of symbols L+1, L+2, ..., L+K indicates that two column of the Alamouti Space-Time Block Coding respective channels matrix H that then each group of transmitting antenna is sent in 2L transmitting antenna are denoted as h_{: m}, m=1,2 ..., L；Two column of the also respective channels matrix H of set of symbols transmitted by each transmitting antenna, are denoted as h in K transmitting antenna_m, m=L+1, L+2 ..., L+K.

The embodiment of the present invention five is provided when the additional K group code of 2L+K transmitting antenna transmitting L group Alamouti Space-Time Block Coding of transmitting terminal, utilizes the received method for receiving signal detection signal of the N number of receiving antenna in receiving end.Wherein, 2L transmitting antenna transmitting passes through the L group Alamouti Space-Time Block Coding obtained after Alamouti Space-Time Block Coding encoder channel coding by L group code, and K transmitting antenna directly emits K group code.This method detects at least two symbols of transmitting terminal transmitting in multiple-input, multiple-output mimo system, wherein at least one symbol at least two symbol is emitted by least one transmitting antenna again after transmitting terminal carries out channel coding by an encoder；The encoder carries out channel coding to input symbol in such a way that symbol is duplicate and obtains channel signal, the channel signal includes a former input symbol, or the negative value of symbol is inputted including one, or the multiple conjugate value of symbol is inputted including one, or the negative multiple conjugate value of symbol is inputted including one, the channel signal of the encoder output, which is emitted by least one transmitting antenna and passes through at least two different channels, reaches receiving end；This method comprises:

A. at least two receiving antennas of receiving end receive the channel signal that transmitting terminal is emitted, and obtain at least two and receive signal；

B. receiving end carries out channel estimation, obtains the channel matrix being made of the channel coefficients between transmitting antenna and receiving antenna；

C. the evaluated error covariance matrix of the partial symbols at least two symbol is calculated using channel matrix, then the evaluated error covariance matrix of the partial symbols and the newly-increased matrix entries obtained according to the evaluated error covariance matrix and channel matrix H of the partial symbols are utilized, recursion is acquired including the partial symbols and number is more than the evaluated error covariance matrix of the symbol of the partial symbols number, the newly-increased matrix entries be include the newly-increased matrix entries of the partial transmitting signal and number more than the evaluated error covariance matrix of the evaluated error covariance matrix partial transmitting signal of the transmitting signal of the partial transmitting signal number；

D. some items in the obtained evaluated error covariance matrix of step c or evaluated error covariance matrix, at least one of at least two symbols of detection transmitting terminal transmitting are utilized.

Fig. 6 show the flow chart for calculating the initial value of evaluated error covariance matrix of all transmitting set of symbols, including the following steps:

Step 601: receiving end receives transmitting terminal after the L group Alamouti Space-Time Block Coding and K group code that 2L+K transmitting antenna emits respectively, obtain N number of reception signal, and channel estimation is carried out according to signal is received, obtain the N being made of the channel coefficients in single symbol period × (2L+K) channel matrix, the extended channel matrices H of 2N × 2 (L+K) in two symbol periods in an Alamouti Space-Time Block Coding period is thus constructed again, the transmitting symbolic vector s in two symbol periods in Alamouti Space-Time Block Coding period is constructed, is respectively as follows:

$s={\left[\begin{array}{cccccccccccccc}{s}_1^1& {\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">s</figure-callout>}}_2^1& {\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">s</figure-callout>}}_1^2& {\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">s</figure-callout>}}_2^2& \&CenterDot;\&CenterDot;\&CenterDot;& s_1^L& s_2^{\mathrm{<figure-callout\; id="1"\; label="L\; b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">L</figure-callout>}}& b_{}^{}{}_1^1& {-\left({\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^1\right)}^{*}& {\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^2& -{\left({\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^2\right)}^{*}& \&CenterDot;\&CenterDot;\&CenterDot;& b_1^K& -{\left(b_2^K\right)}^{*}\end{array}\right]}^T$ It can also be expressed as $s={\left[\begin{array}{cccccccccccccc}{s}_1^1& {\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">s</figure-callout>}}_2^1& {\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">s</figure-callout>}}_1^2& {\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">s</figure-callout>}}_2^2& \&CenterDot;\&CenterDot;\&CenterDot;& s_1^L& s_2^{\mathrm{<figure-callout\; id="1"\; label="L\; b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">L</figure-callout>}}& b_{}^{}{}_1^1& {-\left({\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^1\right)}^{*}& {\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">b</figure-callout>}}_1^2& -{\left({\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">b</figure-callout>}}_2^2\right)}^{*}& \&CenterDot;\&CenterDot;\&CenterDot;& b_1^K& -{\left(b_2^K\right)}^{*}\end{array}\right]}^T$ $={\left[\begin{array}{cccccccccccccc}{s}_1^1& {\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">s</figure-callout>}}_2^1& {\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">s</figure-callout>}}_1^2& {\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">s</figure-callout>}}_2^2& \&CenterDot;\&CenterDot;\&CenterDot;& s_1^L& s_2^L& s_1^{L+1}& {-\left(s_2^{L+1}\right)}^{*}& s_1^{L+2}& -{\left(s_2^{L+2}\right)}^{*}& \&CenterDot;\&CenterDot;\&CenterDot;& s_1^{L+K}& -{\left(s_2^{L+K}\right)}^{*}\end{array}\right]}^T$

At this moment, the received signal vector r in two symbol periods in an Alamouti Space-Time Block Coding period are as follows: $r=\left[\begin{array}{c}{r}_{11}\\ r_{12}^{*}\\ r_{21}\\ r_{22}^{*}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ r_{N1}\\ {\mathrm{<figure-callout\; id="2"\; label="r\; N"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">r</figure-callout>}}_N^2\end{array}\right].$

Here, in actual application, can not have to specifically obtain above-mentioned extended channel matrices, because each channel coefficients in extended channel matrices are made of the channel coefficients in single symbol period.

The sequencing that all transmitting set of symbols 1,2 ..., L, L+1 ..., L+K are detected in receiving end is preset, is denoted as t with transmitting set of symbols serial number_L+K, t_L+K-1..., t_L+1, t_L... t₂, t₁, then, extended channel matrices H is then obtained by column rearrangement accordingly $H_{L+K}^{\left(t_{L+K}\right)}=[{\stackrel{\&OverBar;}{h}}_{{:t}_1}{\stackrel{\&OverBar;}{h}}_{{:\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}\&CenterDot;\&CenterDot;\&CenterDot;{\stackrel{\&OverBar;}{h}}_{{:t}_{L+K-1}}{\stackrel{\&OverBar;}{h}}_{{:t}_{L+K}}],$ Wherein,

Indicate transmitting set of symbols t_mCorresponding two column.

Utilize vector f=[t₁, t₂..., t_m..., t_L+K-1, t_L+K]^TRecord and extended channel matrices

The index of corresponding set of symbols.

Step 602: indicating the total number for the set of symbols that transmitting terminal is emitted with M, that is, set M=L+K, then extended channel matrices

It can be expressed as

Use extended channel matrices

First acquireCross-correlation channel matrix ${\mathrm{\&Phi;}}_M={\left(H_M^{\left(t_M\right)}\right)}^H\&CenterDot;H_M^{\left(t_M\right)},$ Again by Φ_MAcquire estimation error covariance matrix Q_MInverse matrix $R_M={\left(H_M^{\left(t_M\right)}\right)}^H\&CenterDot;H_M^{\left(t_M\right)}+\mathrm{\&alpha;}{I}_{2M\&times;2M}={\mathrm{\&Phi;}}_M+\mathrm{\&alpha;}{I}_{2M\&times;2M}.$

Wherein, obtained R_MAre as follows:

$R_M={\left(H_M^{\left(t_M\right)}\right)}^H\&CenterDot;H_M^{\left(t_M\right)}+\mathrm{\&alpha;}{I}_{2M\&times;2M}$

R_MInIt is all 2 × 2 matrix-block, wherein * expression takes conjugate transposition to 1 matrix.Meanwhile the matrix-block on diagonal line is ${\stackrel{\&OverBar;}{r}}_{t_i{t}_i}=\left[\begin{array}{cc}{r}_{t_i{t}_i}& 0\\ 0& r_{t_i{t}_i}\end{array}\right],$ And the matrix-block on off-diagonal is ${\stackrel{\&OverBar;}{r}}_{t_i{t}_j}=\left[\begin{array}{cc}{r}_{t_i{t}_j}& r_{t_i{t}_j}^{\&prime;}\\ -{\left(r_{t_i{t}_j}^{\&prime;}\right)}^{*}& {\left(r_{t_i{t}_j}\right)}^{*}\end{array}\right].$ So for the matrix-block on diagonal line, it is only necessary to calculate one therein；And for the matrix-block on off-diagonal, it is only necessary to calculate two thereinWith

.(R simultaneously_M)^H=R_M, to only need to calculate R_MAll matrix-blocks of middle diagonal line side.

We are satisfaction $A=\left[\begin{array}{cc}{a}_1& a_2\\ -{\left(a_2\right)}^{*}& {\left(a_1\right)}^{*}\end{array}\right]$ Matrix be known as Alamouti matrix, it is readily seen that above-mentioned matrix-block ${\stackrel{\&OverBar;}{r}}_{t_i{t}_i}=\left[\begin{array}{cc}{r}_{t_i{t}_i}& 0\\ 0& r_{t_i{t}_i}\end{array}\right]$ With ${\stackrel{\&OverBar;}{r}}_{t_i{t}_j}=\left[\begin{array}{cc}{r}_{t_i{t}_j}& r_{t_i{t}_j}^{\&prime;}\\ -{\left(r_{t_i{t}_j}^{\&prime;}\right)}^{*}& {\left(r_{t_i{t}_j}\right)}^{*}\end{array}\right]$ It is all Alamouti matrix, and when needing to calculate an Alamouti matrix, do not need to calculate whole in the matrix 4, it is only necessary to calculate two therein, such as only calculate first row $\left[\begin{array}{c}{a}_1\\ -{\left(a_2\right)}^{*}\end{array}\right],$ Or only calculate the first row [a₁ a₂]。

Step 603: calculating a finally detected transmitting set of symbols t₁Corresponding evaluated error covariance matrix, is denoted as

One transmitting set of symbols t₁Two symbols emitted in two symbol periods in exactly one Alamouti Space-Time Block Coding period by same group of antenna, and the same group of antenna, it is two antennas when emitting Alamouti Space-Time Block Coding, is an antenna when emitting the signal of spatial reuse all the way.

Corresponding transmitting set of symbols t₁Extended channel matrices be $H_1^{\left(t_1\right)}=\left[{\stackrel{\&OverBar;}{h}}_{{:t}_1}\right].$ From the R calculated in step 602_MIn, obtain transmitting set of symbols t₁The inverse matrix of evaluated error covariance matrix be $R_1^{\left(t_1\right)}={\left({\stackrel{\&OverBar;}{h}}_{{:t}_1}\right)}^H\&CenterDot;{\stackrel{\&OverBar;}{h}}_{{:\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}+\mathrm{\&alpha;}{I}_{2\&times;2},$ It is easy

See being exactly R^(M)The 1st row the 1st arranges 2 × 2 matrix-block arranged to the 2nd row the 2nd on diagonal line ${\stackrel{\&OverBar;}{r}}_{t_1{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}=\left[\begin{array}{cc}{r}_{t_1{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}& 0\\ 0& r_{t_1{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">t</figure-callout>}}_1}\end{array}\right].$

By $Q_1^{\left(t_1\right)}={\left(R_1^{\left(t_1\right)}\right)}^{-1}$ It obtains

$Q_1^{\left(t_1\right)}=\left[\begin{array}{cc}{\left(r_{t_1{t}_1}\right)}^{-1}& 0\\ 0& {\left(r_{t_1{t}_1}\right)}^{-1}\end{array}\right].$

The m transmitting set of symbols t that recursion is finally detected below_m..., t₂, t₁Corresponding evaluated error covariance matrix

Firstly, m is allowed to be equal to 2,604 are entered step.

Step 604: judging whether to have obtained the M-1 evaluated error covariance matrixes for being detected transmitting set of symbols, that is, judge whether m is greater than M-1, if it is, having obtained the M-1 evaluated error covariance matrixes for being detected transmitting set of symbols, go to step 608；Otherwise, recursion m are detected the evaluated error covariance matrix for emitting set of symbols

Value, execute step 605,606,607.

Step 605: finally m detected transmitting set of symbols t_m..., t₂, t₁Corresponding extended channel matrices are $H_m^{\left(t_m\right)}=[{\stackrel{\&OverBar;}{h}}_{{:t}_1}{\stackrel{\&OverBar;}{h}}_{{:\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}\&CenterDot;\&CenterDot;\&CenterDot;{\stackrel{\&OverBar;}{h}}_{{:t}_m}],$ Therefore, the inverse matrix of corresponding evaluated error covariance matrix is $R_m^{\left(t_m\right)}={\left(H_m^{\left(t_m\right)}\right)}^H\&CenterDot;H_m^{\left(t_m\right)}+\mathrm{\&alpha;}{I}_{\left(2m\right)\&times;\left(2m\right)}.$

With

There is following recurrence relation:

$R_m^{\left(t_m\right)}=\left[\begin{array}{cc}{R}_{m-1}^{\left(t_{m-1}\right)}& {\stackrel{\&OverBar;}{v}}_{m-1}^{\left(t_m\right)}\\ {\left({\stackrel{\&OverBar;}{v}}_{m-1}^{\left(t_m\right)}\right)}^H& {\stackrel{\&OverBar;}{\mathrm{\&beta;}}}_1^{\left(t_m\right)}\end{array}\right],$ Wherein,

It is the result either initial value of last recursion

${\stackrel{\&OverBar;}{\mathrm{\&beta;}}}_1^{\left(t_m\right)}={{\stackrel{\&OverBar;}{h}}_{{:t}_m}}^H\&CenterDot;{\stackrel{\&OverBar;}{h}}_{{:t}_m}+\mathrm{\&alpha;}{I}_{2\&times;2}={\stackrel{\&OverBar;}{r}}_{t_m{t}_m};$ ${\stackrel{\&OverBar;}{v}}_{m-1}^{\left(t_m\right)}=\left[\begin{array}{c}{{\stackrel{\&OverBar;}{h}}_{{:t}_1}}^H\&CenterDot;{\stackrel{\&OverBar;}{h}}_{{:t}_m}\\ {{\stackrel{\&OverBar;}{h}}_{{:t}_1}}^H\&CenterDot;{\stackrel{\&OverBar;}{h}}_{{:t}_m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {{\stackrel{\&OverBar;}{h}}_{{:t}_{m-1}}}^H\&CenterDot;{\stackrel{\&OverBar;}{h}}_{{:t}_m}\end{array}\right]=\left[\begin{array}{c}{\stackrel{\&OverBar;}{r}}_{t_1{t}_m}\\ {\stackrel{\&OverBar;}{r}}_{t_2{t}_m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {\stackrel{\&OverBar;}{r}}_{t_{m-1}{t}_m}\end{array}\right].$

It is readily seen that

With

The R that can be calculated from step 602_MIn directly obtain, more specifically,

It is R _M2m-1 row 2m-1 arranges 2 × 2 matrix-block arranged to 2m row 2m on diagonal line, and

It is by R_MThe matrix-block of 2 (m-1) rows 2 column of head 2 (m-1) the row composition of 2m-1 column and 2m column.Without any calculating, so that it may directly obtain

It can easily be seen that ${\stackrel{\&OverBar;}{\mathrm{\&beta;}}}_1^{\left(t_m\right)}={\stackrel{\&OverBar;}{r}}_{t_m{t}_m}=\left[\begin{array}{cc}{r}_{t_{\mathrm{<figure-callout\; id="0"\; label="m\; t\; m"\; filenames="H2008101874159E00101.png"\; state="\{\{state\}\}">m</figure-callout>}}{t}_m{}_{}}& 0\\ 0& r_{t_m{t}_m}\end{array}\right],$ Thus by

The item of the first row first row can be obtained by entirely

The item of the first row first row is denoted as

Similarly, byFirst row can directly obtain its secondary series,

First row be denoted as

Step 606: seeking m finally detected transmitting set of symbols t_m..., t₂, t₁Corresponding evaluated error covariance matrix

Utilize the evaluated error covariance matrix of a corresponding finally detected transmitting set of symbols

Or the m-1 transmitting set of symbols t that the obtained correspondence of last recursion is finally detected_m-1..., t₂, t₁Evaluated error covariance matrixRecursion obtains

Recurrence method is as described below:

First by

With

It calculates ${\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}=-Q_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}$ With ${\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}-{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{Q}_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}};$ Using the symmetry of Alamouti matrix, can directly by

It obtainsWithout calculating, i.e., ${\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}=\left[\begin{array}{c}{\mathrm{\&mu;}}_1\\ {\mathrm{\&mu;}}_2\\ {\mathrm{\&mu;}}_3\\ {\mathrm{\&mu;}}_4\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {\mathrm{\&mu;}}_{2m-3}\\ {\mathrm{\&mu;}}_{2m-2}\end{array}\right],$ SoNecessarily satisfying for ${\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}=\left[\begin{array}{c}-{\left({\mathrm{\&mu;}}_2\right)}^{*}\\ {\left({\mathrm{\&mu;}}_1\right)}^{*}\\ -{\left({\mathrm{\&mu;}}_4\right)}^{*}\\ {\left({\mathrm{\&mu;}}_3\right)}^{*}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ -{\left({\mathrm{\&mu;}}_{2m-2}\right)}^{*}\\ {\left({\mathrm{\&mu;}}_{2m-3}\right)}^{*}\end{array}\right],$ To

Each element can directly by

Each element obtain, pass through some operations for taking negative sign and taking conjugation.

Then by

With

Pass through $T_{m-1}^{\left(t_m\right)}=Q_{m-1}^{\left(t_m\right)}+{\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H+{\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}\right)}^H)$ It calculates

Finally by $Q_m^{\left(t_m\right)}=\left[\begin{array}{ccc}{T}_{m-1}^{\left(t_{m-1}\right)}& {\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}& {\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}\\ {\left({\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H& {\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}& 0\\ {\left({\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}\right)}^H& 0& {\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}\end{array}\right]$ It obtainsPay attention in above-mentioned calculating process, matrix T_m-1、Q_m-1、 ${\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H,{\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}\right)}^H$ With ${\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H+{\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}\right)}^H)$ All be about diagonal line Hermitian it is symmetrical (i.e. matrix meet the matrix conjugate transposition it is equal with the matrix itself), to only need to calculate all items of above-mentioned each diagonal of a matrix side and save calculation amount；The item arranged with (x, y) representing matrix xth row y, then by (2i-1 in these matrixes, 2j-1), (2i-1,2j), (2i, 2j-1), (2i, 2j) submatrix of 2 rows 2 column of this 4 compositions is all Alamouti matrix, here i=1,2, ..., m-1, j=1,2 ..., m-1, to in the submatrix of each 2 row 2 column, it is only necessary to item i.e. 2 of half are calculated, so as to further save calculation amount；

Do not need to calculate yet, can directly by

It obtains.

It is above-mentioned $T_{m-1}^{\left(t_m\right)}=Q_{m-1}^{\left(t_m\right)}+{\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H+{\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}\right)}^H)$ Referred to as

Broad sense thank to Germania-Morrison (Sherman-Morrison) as a result, because ${\underset{\&OverBar;}{T}}_{m-1}^{\left(t_m\right)}=Q_{m-1}^{\left(t_m\right)}+{\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}\right)}^H$ Meet above-mentioned

The definition for thanking to Germania-Morrison (Sherman-Morrison) result, and $T_{m-1}^{\left(t_m\right)}={\underset{\&OverBar;}{T}}_{m-1}^{\left(t_m\right)}+{\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}{\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}{\left({\mathrm{\&mu;}}_{m-1}^{\&prime;\left(t_m\right)}\right)}^H$ It is equivalent to handle

Regard Q matrix as, then asks and once thank to Germania-Morrison (Sherman-Morrison) result.

And when specific implementation, in order to reduce calculation amount, it is above-mentioned by

It obtainsRecurrence method in, calculate ${\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}-{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{Q}_{m-1}^{\left(t_{m-1}\right)}{v}_{m-1}^{\left(t_m\right)}}$ The step of can be refined as ${\mathrm{\&omega;}}_{m-1}^{\left(t_m\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_m\right)}-{\left(v_{m-1}^{\left(t_m\right)}\right)}^H{\mathrm{\&mu;}}_{m-1}^{\left(t_m\right)}}.$

The value of step 607:m increases by 1, i.e. m=m+1, then return step 604.

Step 608: having obtained M-1 transmitting set of symbols t now_M-1..., t_m..., t₂, t₁Corresponding evaluated error covariance matrix

Value.First by

WithIt calculates ${\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}=-Q_{M-1}^{\left(t_{M-1}\right)}{v}_{M-1}^{\left(t_M\right)}$ With ${\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_M\right)}-{\left(v_{M-1}^{\left(t_M\right)}\right)}^H{Q}_{M-1}^{\left(t_{M-1}\right)}{v}_{M-1}^{\left(t_M\right)}};$ When specific implementation, in order to reduce calculation amount, above-mentioned calculating process can be refined are as follows: first byWith

It calculates ${\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}=-Q_{M-1}^{\left(t_{M-1}\right)}{v}_{M-1}^{\left(t_M\right)},$ Again byIt calculates ${\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}=\frac{1}{{\mathrm{\&beta;}}_1^{\left(t_M\right)}+{\left(v_{M-1}^{\left(t_M\right)}\right)}^H{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}}.$

Then, using the method described in step 606, by

It directly obtains

Again by

With

It calculates

Diagonal entry, i.e., $T_{M-1}^{\left(t_M\right)}(i,i)=Q_{M-1}^{\left(t_{M-1}\right)}(i,i)+{\mathrm{\&omega;}}_{M-1}^{\left(t_{M-1}\right)}({\mathrm{\&mu;}}_{M-1}^{\left(t_{M-1}\right)}\left(i\right){\left({\mathrm{\&mu;}}_{M-1}^{\left(t_{M-1}\right)}\left(i\right)\right)}^H+{\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_{M-1}\right)}\left(i\right){\left({\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_{M-1}\right)}\left(i\right)\right)}^H),$

Wherein i=1,2 ..., 2 (M-1),With

Respectively indicate matrix

WithI-th of element on diagonal line, i.e. matrix

With

Element on i-th row i-th column, and

With

Indicate column vectorWith

I-th.Because having $T_{M-1}^{\left(t_M\right)}(2k-\mathrm{1,2}k-1)=T_{M-1}^{\left(t_M\right)}(2k,2k),$ K=1,2 ..., (M-1), so actually we only need to calculate $T_{M-1}^{\left(t_M\right)}(i,i)=Q_{M-1}^{\left(t_{M-1}\right)}(i,i)+{\mathrm{\&omega;}}_{M-1}^{\left(t_{M-1}\right)}({\mathrm{\&mu;}}_{M-1}^{\left(t_{M-1}\right)}\left(i\right){\left({\mathrm{\&mu;}}_{M-1}^{\left(t_{M-1}\right)}\left(i\right)\right)}^H+{\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_{M-1}\right)}\left(i\right){\left({\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_{M-1}\right)}\left(i\right)\right)}^H),$ I=1,3,5,7..., 2M-3, so that it may acquire

All items on diagonal line.

M=M is enabled in expression formula above-mentioned, we obtain expression formula $Q_M^{\left(t_M\right)}=\left[\begin{array}{ccc}{T}_{M-1}^{\left(t_{M-1}\right)}& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_M\right)}\\ {\left({\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\right)}^H& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}& 0\\ {\left({\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_M\right)}\right)}^{\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="H2008101874159E00101.png"\; state="\{\{state\}\}">H</figure-callout>}}& 0& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}\end{array}\right],$ We do not need to calculate this expression formula, and only by this expression formula, we be can be clearly seen that, by the above-mentioned calculating process of this step, matrix is had been obtained in we

Diagonal line on all elements, and they are exactly

And matrix

Diagonal entryI=1,2 ..., 2 (M-1).

Because the set of all M transmitting set of symbols be it is identical, under the scene of ideal Floating-point Computation, all M evaluated error covariance matrixes for emitting set of symbols are unique；And the sequencing t that pre-set all M transmitting set of symbols are detected in receiving end_M, t_M-1..., t_m..., t₂, t₁Difference only results in eachCorresponding exchange is done in the position of interior row and column, and each

The element set for inside including is identical.So we can remember $Q_M^{\left(t_M\right)}=Q_M.$

The step of commencing signal detects below notices that we have acquired the evaluated error covariance matrix of M all transmitting set of symbols $Q_M=Q_M^{\left(t_M\right)}$ Diagonal line on all elements, andIt is exactly that optimal detection sequence is determined in signal detection process, and when gradually detecting according to the optimal detection sequence and using interference elimination method each transmitting set of symbols, the initial value of used evaluated error covariance matrix.Furthermore we acquire during above-mentioned recursion

With it is each

M=1,2 ..., M-1, described is each

It is the evaluated error covariance matrix of part transmitting set of symbols to be detected, is corresponding transmitting set of symbols t_m..., t₂, t₁Evaluated error covariance matrix.

After obtaining the initial value of the evaluated error covariance matrix of all transmitting set of symbols to be detected, into the process of detection signal shown in Fig. 7, that is, a of Fig. 7 is gone to.

Fig. 7 is the flow chart of signal detection, and signal detection shown in Fig. 7 is since a.

Step 701: the index for emitting set of symbols is still vector f=[t₁, t₂..., t_m..., t_M-1, t_M]^T.Initial value for the Q of iteration in signal detection process is denoted as $Q_M=Q_M^{\left(t_M\right)};$ P^(M)/2Corresponding extended channel matrices are exactly

It is denoted as $H_M=H_M^{\left(t_M\right)};$ And the index for emitting set of symbols accordingly is still vector f=[t₁, t₂..., t_m..., t_M-1, t_M]^T.To the signal received $r={[r_{11},r_{12}^{*},r_{21},r_{22}^{*},\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{<figure-callout\; id="1"\; label="r\; N"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">r</figure-callout>}}_N,{\mathrm{<figure-callout\; id="2"\; label="r\; N"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">r</figure-callout>}}_N^2]}^T$ Pre-matching filtering transformation is carried out, the pre-matching filter result z of received signal vector r is obtained_M=(H^(M))^HR, wherein (H^(M))^HFor matched filter.The iteration variable m that will test in signal process below is set as 1, enters step 702.

Step 702: emitting in set of symbols at M and determine the best transmitting set of symbols of received signal to noise ratio, method is: in Q_MDiagonal entry in, search the taken the smallest item of value, be denoted as l_MCapable and l_MThe item of column, i.e., $l_M=\underset{i=\mathrm{1,3,5}...}{\overset{2M-1}{\arg \min }}{Q}_{M-m+1}(i,i),$ Pay attention to i=2k-1, k=1,2 ..., M.The Q_M+1-mL in diagonal entry_MCapable and l_MThe item of column corresponds to the transmitting set of symbols that received signal to noise ratio is best in M+1-m transmitting set of symbols, i.e. currently detected transmitting set of symbols.

Step 703: calculated for subsequent signal detection needs the matrix Q used_MOff-diagonal element, they are exactly Q_ML_MCapable off-diagonal element and l_MThe off-diagonal element and Q of column_ML_M+1Capable off-diagonal element and l_M+1The off-diagonal element of column.According to matrix Q_MMeet $Q_M^H=Q_M$ Symmetric relation to Q_M(i, j)=(Q_M(j, i))^*, it is only necessary to calculate Q_ML_MThe off-diagonal element Q of column_M(i, l_M), i=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., 2M and l_MThe off-diagonal element Q of+1 column_M(i, l_M+ 1), i=1,2 ..., l_M, l_M+ 2, l_M+ 3 ..., 2M.And Q_ML_MCapable each off-diagonal element Q_M(l_M, i) and (i=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., M) it can be by Q_ML_MThe element Q of the corresponding position of column_M(i, l_M) conjugation is taken to obtain；Q_ML_MEach off-diagonal element Q of+1 row_M(l_M+ 1, i) (i=1,2 ..., l_M, l_M+ 2, l_M+ 3 ..., M) it can be by Q_ML_MThe element Q of the corresponding position of+1 column_M(i, l_M+ 1) conjugation is taken to obtain.

And Q_ML_MColumn and l_MThe off-diagonal element of+1 column, by aforementioned $T_{M-1}^{\left(t_M\right)}{=Q}_{M-1}^{\left(t_{M-1}\right)}+{\mathrm{\&omega;}}_{M-1}^{\left(t_{M-1}\right)}({\mathrm{\&mu;}}_{M-1}^{\left(t_{M-1}\right)}{\left({\mathrm{\&mu;}}_{M-1}^{\left(t_{M-1}\right)}\right)}^H+{\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_{M-1}\right)}{\left({\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_{M-1}\right)}\right)}^H)$ With $Q_M^{\left(t_M\right)}=\left[\begin{array}{ccc}{T}_{M-1}^{\left(t_{M-1}\right)}& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_M\right)}\\ {\left({\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\right)}^H& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}& 0\\ {\left({\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}{\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_M\right)}\right)}^{\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="H2008101874159E00101.png"\; state="\{\{state\}\}">H</figure-callout>}}& 0& {\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}\end{array}\right]$ It calculates.

Specifically, if l_M=2M-1, then Q_M2M-1 column off diagonal element be

With 0, corresponding Q_M2M column off diagonal element be

With 0, pay attention to

Do not need to calculate, can directly by

It obtains.And if l_M≠ 2M-1, then Q_ML_MThe off diagonal element of column respectively by

$Q_M^{\left(t_M\right)}(i,l_M)=T_{M-1}^{\left(t_M\right)}(i,l_M)=Q_{M-1}^{\left(t_{M-1}\right)}(i,l_M)+{\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}({\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\left(i\right){\left({\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\left(l_M\right)\right)}^{*}+{\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_M\right)}\left(i\right){\left({\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_M\right)}\left(l_M\right)\right)}^{*}),$ I=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., 2 (M-1), and $Q_M^{\left(t_M\right)}(2M-1,l_M)={\left({\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}\right)}^{*}{\left({\mathrm{\&mu;}}_{M-1}^{\left(t_M\right)}\left(l_M\right)\right)}^{*}$ With $Q_M^{\left(t_M\right)}(2M,l_M)={\left({\mathrm{\&omega;}}_{M-1}^{\left(t_M\right)}\right)}^{*}{\left({\mathrm{\&mu;}}_{M-1}^{\&prime;\left(t_M\right)}\left(l_M\right)\right)}^{*}$ It calculates；And Q_ML_MThe off diagonal element of+1 column can be by Q_ML_MThe off diagonal element of column is directly obtained without calculating, and utilizes Q_ML_MColumn and l_MBy (2i-1, l in+1 column_M), (2i-1, l_M+ 1), (2i, l_M), (2i, l_M+ 1) submatrix of 2 rows 2 column of this 4 compositions is all Alamouti matrix, here i=1,2 ..., M.

Step 704: in matrix Q_MMiddle exchange l_MCapable and 2M-1 row exchanges l_MColumn and 2M-1 column, then exchange l_M+ 1 row and 2M row exchange l_M+ 1 column and 2M column；Correspondingly, in matrix Φ_MMiddle exchange l_MCapable and 2M-1 row exchanges l_MColumn and 2M-1 column, then exchange l_M+ 1 row and 2M row exchange l_M+ 1 column and 2M column；Correspondingly, to the pre-matching filter result vector z for receiving signal_MMiddle exchange l_MAnd 2M-1, then exchange l_M+ 1 and 2M.(l is exchanged in vector f_M+ 1)/2 and M.Pay attention to matrix Q_MIn 2M-1 row, 2M-1 column, 2M row and 2M column off-diagonal element all do not found out to come in most cases, but this is not interfered in Q_MThe middle exchange for carrying out above-mentioned row and column.

Step 705: currently detected transmitting set of symbols is to be denoted as p M in vector f_m, p_m=f (M).The estimated value for calculating two symbols to currently detected transmitting set of symbols in two symbol periods, if what is be currently detected is to carry out the set of symbols of Alamouti Space-Time Block Coding coding $\left[\begin{array}{c}{\tilde{s}}_1^{p_m}\\ {\tilde{s}}_2^{p_m}\end{array}\right]=\left[\begin{array}{c}{q}_M^H\\ q_M^{\&prime;H}\end{array}\right]z_M,$ If what is be currently detected is not carry out the set of symbols of Alamouti Space-Time Block Coding coding $\left[\begin{array}{c}{\tilde{s}}_1^{p_m}\\ -{\left({\tilde{s}}_2^{p_m}\right)}^{*}\end{array}\right]=\left[\begin{array}{c}{q}_M^H\\ q_M^{\&prime;H}\end{array}\right]z_M,$ The estimated value to two symbols in transmitting set of symbols is thus obtained again

With

Here q_MIndicate Q_M2M-1 column and q '_M ^HIndicate Q_M2M column.

Step 706: to the estimated value of obtained transmitting set of symbols $\left[\begin{array}{c}{\tilde{s}}_1^{p_m}\\ {\tilde{s}}_2^{p_m}\end{array}\right]$ Quantified, obtains the testing result to transmitting set of symbols $\left[\begin{array}{c}{\hat{s}}_1^{p_m}\\ {\hat{s}}_2^{p_m}\end{array}\right].$

Step 707: from the influence for eliminating two symbols in currently detected transmitting set of symbols in the pre-matching filter result of received signal vector, the corresponding multiple pre-matching filter results for receiving signal of M-m transmitting set of symbols being not yet detected, to which signal detection problem next time to be become to the detection of M-m transmitting set of symbols by interference cancellation techniques, specific method is: deleting the column vector z for having 2M_MLast 2 obtain the column vector (z of 2 (M-1) items_M)^min us；From (z_M)^min usThe middle interference for eliminating two symbols in the transmitting set of symbols being currently detected obtains if currently detected is the set of symbols for carrying out Alamouti Space-Time Block Coding coding $z_{M-1}={\left(z_M\right)}^{\min \mathrm{us}}-{\mathrm{\&phi;}}_M\&CenterDot;\left[\begin{array}{c}{\hat{s}}_1^{p_m}\\ {\hat{s}}_2^{p_m}\end{array}\right];$ If currently detected is the set of symbols for not carrying out Alamouti Space-Time Block Coding coding, obtain $z_{M-1}={\left(z_M\right)}^{\min \mathrm{us}}-{\mathrm{\&phi;}}_M\&CenterDot;\left[\begin{array}{c}{\hat{s}}_1^{p_m}\\ -{\left({\hat{s}}_2^{p_m}\right)}^{*}\end{array}\right].$ Wherein φ_MIt is matrix Φ^(M)It is last 2 column be 2M-1 column and 2M column head 2M-2 row.

Step 708: from matrix Φ_M+1-mMiddle deletion 2M column and 2M row, then 2M-1 column and 2M-1 row are deleted, obtain matrix Φ_M-m。

Start the 2nd below, 3 ... until the detection of m-th transmitting set of symbols, into Fig. 8.

Step 801: judging shown in Fig. 2 to seek Q_MInitial value during, if obtained the corresponding M-m evaluated error covariance matrix Q for emitting set of symbols_M-m, if so, thening follow the steps 802；Otherwise, step 803 is executed.

Step 802: by recursion Q_MInitial value during obtained Q_M-mAs the evaluated error covariance matrix Q for corresponding to M-m transmitting set of symbols required for next iteration_M-m.Then step 807 is executed.

We are available $Q_{M+1-m}=\left[\begin{array}{ccc}{T}_{M-m}& w_{M-m}& w_{M-m}^{\&prime;}\\ {\left(w_{M-m}\right)}^H& {\mathrm{\&omega;}}_{M-m}& 0\\ {\left(w_{M-m}^{\&prime;}\right)}^{\mathrm{<figure-callout\; id="0"\; label="H"\; filenames="H2008101874159E00101.png"\; state="\{\{state\}\}">H</figure-callout>}}& 0& {\mathrm{\&omega;}}_{M-m}\end{array}\right],$ Wherein w '_M-mIt can be directly by w_M-mIt obtains, because of column vector w_M-mWith w '_M-mMatrix [the w of composition_M-m w′_M-m] be necessarily made of completely the Alamouti submatrix that several 2 rows 2 arrange；And by

$Q_{M-m}=T_{M-m}-\frac{1}{{\mathrm{\&omega;}}_{M-m}}(w_{M-m}\&CenterDot;{\left(w_{M-m}\right)}^H+w_{M-m}^{\&prime;}\&CenterDot;{\left(w_{M-m}^{\&prime;}\right)}^H)$ Q can be calculated_M-m.And the embodiment of the present invention only utilizes matrix Q_M+1-mCalculate the evaluated error covariance matrix Q that M-m transmitting set of symbols is corresponded to required for next iteration_M-mIn required item, specific method is:

Step 803: first by T_M-m、w_M-mWith w '_M-mCalculate Q_M-mDiagonal entry, i.e.,

$Q_{M-m}(i,i)=T_{M-m}(i,i)-\frac{1}{{\mathrm{\&omega;}}_{M-m}}(w_{M-m}\left(i\right)\&CenterDot;{\left(w_{M-m}\left(i\right)\right)}^H+w_{M-m}^{\&prime;}\left(i\right)\&CenterDot;{\left(w_{M-m}^{\&prime;}\left(i\right)\right)}^H),$ Wherein i=1,2 ..., 2 (M-m), Q_M-m(i, i) and T_M-m(i, i) respectively indicates matrix Q_M-mAnd T_M-mI-th of element on diagonal line, i.e. matrix Q_M-mAnd T_M-mElement on i-th row i-th column, and w_M-m(i) and w '_M-m(i) column vector w is respectively indicated_M-mWith w '_M-mI-th.And because there is Q_M-m(2k-1,2k-1)=Q_M-m(2k, 2k), k=1,2 ..., (M-m), so actually we only need to calculate

$Q_{M-m}(i,i)=T_{M-m}(i,i)-\frac{1}{{\mathrm{\&omega;}}_{M-m}}(w_{M-m}\left(i\right)\&CenterDot;{\left(w_{M-m}\left(i\right)\right)}^H+w_{M-m}^{\&prime;}\left(i\right)\&CenterDot;{\left(w_{M-m}^{\&prime;}\left(i\right)\right)}^H),$ I=1,3,5,7..., 2 (M-m) -1, so that it may acquire Q_M-mAll items on diagonal line.

Step 804: determining the best transmitting set of symbols of received signal to noise ratio in M-m transmitting set of symbols to be detected, method is: in Q_M-mDiagonal entry in, search the taken the smallest item of value, be denoted as l_M-mCapable and l_M-mThe item of column, i.e., $l_{M-m}=\underset{i=\mathrm{1,3,5}...}{\overset{2(M-m)-1}{\arg \min }}{Q}_{M-m}(i,i),$ Pay attention to i=2k-1, k=1,2 ..., M-m.The Q_M-mL in diagonal entry_M-mCapable and l_M-mThe item of column corresponds to the transmitting set of symbols that received signal to noise ratio is best in M-m transmitting set of symbols, i.e. currently detected transmitting set of symbols.

Step 805: calculated for subsequent signal detection needs the matrix Q used_M-mOff-diagonal element, they are exactly Q_M-mL_M-mCapable off-diagonal element and l_M-mThe off-diagonal element and l of column_M-mThe off-diagonal element and l of+1 row_M-mThe off-diagonal element of+1 column.According to matrix Q_M-mMeet $Q_{M-m}^H=Q_{M-m}$ Symmetric relation to Q_M-m(i, j)=(Q_M-m(j, i))^*, it is only necessary to calculate Q_M-mL_M-mThe off-diagonal element Q of column_M-m(i, l_M-m), i=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., 2 (M-m) and l_M-mThe off-diagonal element Q of+1 column_M-m(i, l_M-m+ 1), i=1,2 ..., l_M, l_M+ 2, l_M+ 3 ..., 2 (M-m)；And Q_M-mL_M-mCapable and l_M-mEach off-diagonal element of+1 row can be by Q_M-mL_M-mColumn and l_M-mThe element of the corresponding position of+1 column takes conjugation to obtain.

And Q_M-mL_M-mThe off-diagonal element of column, by

$Q_{M-m}(i,l_{M-m})=T_{M-m}(i,l_{M-m})-\frac{1}{{\mathrm{\&omega;}}_{M-m}}(w_{M-m}\left(i\right)\&CenterDot;{\left(w_{M-m}\left(l_{M-m}\right)\right)}^{*}+w_{M-m}^{\&prime;}\left(i\right)\&CenterDot;{\left(w_{M-m}^{\&prime;}\left(l_{M-m}\right)\right)}^{*})$ It calculates, wherein i=1,2 ..., l_M- 1, l_M+ 1, l_M+ 2 ..., 2 (M-m).And Q_M-mL_M-mThe off diagonal element of+1 column can be by Q_M-mL_M-mThe off diagonal element of column is directly obtained without calculating, and utilizes Q_M-mL_M-mColumn and l_M-mBy (2i-1, l in+1 column_M-m), (2i-1, l_M-m+ 1), (2i, l_M-m), (2i, l_M-m+ 1) submatrix of 2 rows 2 column of this 4 compositions is all Alamouti matrix, here i=1,2 ..., M-m.And T_M-m(i, l_M-m) it is matrix Q_M-m+1In one, if Q_M-m+1It is known that then directly from Q_M-m+1Obtain T_M-m(i, l_M-m)；Otherwise Q is acquired with following methods_M-m+1In this value, such as successively look for Q_M-m+2、Q_M-m+3、...、Q_MIn respective items whether it is known that until in first Q_M-m+k, it is known for finding in k=2 ..., m, then thus according to shown in this step by Q_i+1Item calculate item calculate Q_iRespective items method, by above-mentioned Q_M-m+kA respective items, successively acquire matrix Q_M-m+k-k1, respective items in k1=1,2 ..., k-1 finally acquire Q_M-m+1In respective items be exactly required T_M-m(i, l_M-m)；If until Q_MIn respective items be not still it is known, then the method shown in step 703 acquires Q_MIn respective items value, then thus according to shown in this step by Q_i+1Item calculate item calculate Q_iRespective items method, by above-mentioned Q_MA respective items, successively acquire matrix Q_M-k2, respective items in k2=1,2 ..., m-1 finally acquire Q_M-m+1In respective items be exactly required T_M-m(i, l_M-m)。

Step 806: in matrix Q_M-mMiddle exchange l_M-mCapable and 2 (M-m) -1 rows exchange l_M-mColumn and 2 (M-m) -1 column, then exchange l_M-m+ 1 row and 2 (M-m) row, exchange l_M-m+ 1 column and 2 (M-m) column；Correspondingly, in matrix Φ_M-mMiddle exchange l_M-mCapable and 2 (M-m) -1 rows exchange l_M-mColumn and 2 (M-m) -1 column, then exchange l_M-m+ 1 row and 2 (M-m) row, exchange l_M-m+ 1 column and 2 (M-m) column；Correspondingly, to the pre-matching filter result vector z for receiving signal_M-mMiddle exchange l_M-mItem and the 2nd (M-m) -1, then exchange l_M-m+ 1 and the 2nd (M-m) item.L is exchanged in vector f_M-mAnd M-m.

Similar to the principle in step 704, matrix Q_M-mIn the the 2nd (M-m) -1 row, the 2nd (M-m) -1 column, the 2nd (M-m) be capable and the off-diagonal element of the 2nd (M-m) column is not all found out to come in most cases, but this is not interfered in Q_M-mThe middle exchange for carrying out above-mentioned row and column.

Step 807: currently detected transmitting set of symbols is to be denoted as p M-m in vector f_m+1, p_m+1=f (M-m).The estimated value for calculating two symbols to currently detected transmitting set of symbols in two symbol periods, if what is be currently detected is to carry out the set of symbols of Alamouti Space-Time Block Coding coding $\left[\begin{array}{c}{\tilde{s}}_1^{p_{m+1}}\\ {\tilde{s}}_2^{p_{m+1}}\end{array}\right]=\left[\begin{array}{c}{q}_{M-m}^H\\ q_{M-m}^{\&prime;H}\end{array}\right]z_{M-m},$ If what is be currently detected is not carry out the set of symbols of Alamouti Space-Time Block Coding coding $\left[\begin{array}{c}{\tilde{s}}_1^{p_{m+1}}\\ -{\left({\tilde{s}}_2^{p_{m+1}}\right)}^{*}\end{array}\right]=\left[\begin{array}{c}{q}_{M-m}^H\\ q_{M-m}^{\&prime;H}\end{array}\right]z_{M-m},$ The estimated value to two symbols in transmitting set of symbols is thus obtained again

With

Here q_M-mIndicate Q_M-mThe 2nd (M-m) -1 column and q '_M-m ^HIndicate Q_M-mThe 2nd (M-m) column.

Step 808: to the estimated value of obtained transmitting set of symbols

With

Quantified, obtains the testing result to transmitting set of symbols

With

Step 809: from the influence for eliminating two symbols in currently detected transmitting set of symbols in the pre-matching filter result of received signal vector, the corresponding multiple pre-matching filter results for receiving signal of M-m-1 transmitting set of symbols being not yet detected, to which signal detection problem next time to be become to the detection of M-m transmitting set of symbols by interference cancellation techniques, specific method is: deleting the column vector z for having 2 (M-m) items_M-mLast 2 obtain the column vector (z of 2 (M-m-1) items_M-m)^min us；From (z_M-m)^min usThe middle interference for eliminating two symbols in the transmitting set of symbols being currently detected obtains if currently detected is the set of symbols for carrying out Alamouti Space-Time Block Coding coding $z_{M-m-1}={\left(z_{M-m}\right)}^{\min \mathrm{us}}-{\mathrm{\&phi;}}_{M-m}\&CenterDot;\left[\begin{array}{c}{\hat{s}}_1^{p_{m+1}}\\ {\hat{s}}_2^{p_{m+1}}\end{array}\right];$ If currently detected is the set of symbols for not carrying out Alamouti Space-Time Block Coding coding, obtain $z_{M-m-1}={\left(z_{M-m}\right)}^{\min \mathrm{us}}-{\mathrm{\&phi;}}_{M-m}\&CenterDot;\left[\begin{array}{c}{\hat{s}}_1^{p_{m+1}}\\ -{\left({\hat{s}}_2^{p_{m+1}}\right)}^{*}\end{array}\right].$ Wherein φ_M-mIt is matrix Φ_M-mIt is last 2 column be the 2nd (M-m) -1 column and the 2nd (M-m) column head 2 (M-m-1) row.

Step 810: from matrix Φ_M-mThe 2nd (M-m) of middle deletion column and the 2nd (M-m) row, then the 2nd (M-m) -1 column and the the 2nd (M-m) -1 row are deleted, obtain matrix Φ_M-m-1。

Step 811: judging whether be in next step the last one transmitting set of symbols of detection, that is, judge whether m is equal to M-2, if so, thening follow the steps 813；It is no to then follow the steps 812.

The value increase by 1 of step 812:m, i.e. m=m+1, return step 801.

Step 813: currently detected transmitting set of symbols is the first item of vector f, is denoted as p_M, p_M=f (1).The estimated value for calculating two symbol of the finally detected transmitting set of symbols in two symbol periods, if what is be currently detected is to carry out the set of symbols of Alamouti Space-Time Block Coding coding $\left[\begin{array}{c}{\tilde{s}}_1^{p_M}\\ {\tilde{s}}_2^{p_M}\end{array}\right]=\left[\begin{array}{c}{q}_1^{\mathrm{<figure-callout\; id="1"\; label="H\; q"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">H</figure-callout>}}& \\ q_{}^{}{}_1^{\&prime;\mathrm{<figure-callout\; id="1"\; label="H\; z"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">H</figure-callout>}}& \\ \end{array},\right]z_{}{}_1,$ If what is be currently detected is not carry out the set of symbols of Alamouti Space-Time Block Coding coding $\left[\begin{array}{c}{\tilde{s}}_1^{p_M}\\ -{\left({\tilde{s}}_2^{p_M}\right)}^{*}\end{array}\right]=\left[\begin{array}{c}{q}_1^{\mathrm{<figure-callout\; id="1"\; label="H\; q"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">H</figure-callout>}}& \\ q_{}^{}{}_1^{\&prime;\mathrm{<figure-callout\; id="1"\; label="H\; z"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">H</figure-callout>}}& \\ \end{array},\right]z_{}{}_1,$ The estimated value to two symbols in transmitting set of symbols is thus obtained again

WithHere q₁Indicate Q₁The 1st column and q '₁ ^HIndicate Q₁The 2nd column.

Step 814: to the estimated value of obtained transmitting set of symbols

With

Quantified, obtains the testing result to transmitting set of symbolsWith

By above method, the sequencing detected to transmitting set of symbols is p₁, p₂..., p_M, corresponding, the testing result of M transmitting set of symbols is ${\hat{s}}_1^{{\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">p</figure-callout>}}_1},{\hat{s}}_2^{{\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">p</figure-callout>}}_1},{\hat{s}}_1^{{\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">p</figure-callout>}}_2},{\hat{s}}_2^{{\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">p</figure-callout>}}_2},\&CenterDot;\&CenterDot;\&CenterDot;,{\hat{s}}_1^{p_M},{\hat{s}}_2^{p_M}.$

Method in the embodiment of the present invention one and two is generalized to by the embodiment of the present invention to be emitted in signal by the multichannel that spatial reuse emits comprising at least scene of Alamouti Space-Time Block Coding (STBC) all the way, the symmetry of Alamouti Space-Time Block Coding is further utilized, reduce required calculation amount, and then reduces the time delay for using and reducing signal detection of the hardware resource of receiver.It should be understood that the embodiment of the present invention one to embodiment five signal detection method, can be applied to interference cancellation receiver, linear receiver can also be applied to.

In system shown in FIG. 1, emits and meet relational expression x (k)=Hs (k)+w between the vector of signal and the vector of reception signal, and wherein channel matrix H is the complex matrix of a N × M.In the case where N=M, the inverse matrix H of H can be directly acquired^-1, then by $\hat{s}\left(k\right)=H^{-1}x\left(k\right),$ Obtain the estimated value of emission signal vector

The estimated value

The linear force zero estimated value of referred to as s (k), because by x (k)=Hs (k)+w, it can be seen that theoretically $\hat{s}\left(k\right)=H^{-1}x\left(k\right)=H^{-1}(\mathrm{Hs}\left(k\right)+w)=s\left(k\right)+H^{-1}w.$ The acquisition methods of linear force zero estimated value mentioned herein, namely a kind of method of linear force zero estimated value detection, it should be understood that, it is also a kind of method of signal detection, it is specifically described in the embodiment of the present invention six and embodiment seven, the method for the signal detection can be applied to linear receiver.

The embodiment of the present invention six gives a kind of method of signal detection, the at least two transmitting signal of detection in multiple-input, multiple-output mimo system, the transmitting signal is emitted respectively by each different transmitting antenna of transmitting terminal and reaches receiving end by a channel, and this method includes

A. at least two receiving antennas of receiving end receive the transmitting signal, obtain at least two and receive signal；

B. receiving end carries out channel estimation according to signal is received, and obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna；

C. a sub- inverse of a matrix matrix of matrix H is calculated using channel matrix H, then a sub- inverse of a matrix matrix of calculated matrix H and the newly-increased matrix entries obtained according to channel matrix H are utilized, recursion acquires the inverse matrix of a submatrix comprising the H and a bigger submatrix of the H of a submatrix greater than the H, and the newly-increased matrix entries are the newly-increased matrix entries of a submatrix comprising the H and a submatrix of a bigger submatrix inverse matrix H of the H of a submatrix greater than the H；

D. using the inverse matrix of a bigger submatrix of the obtained H of step c, transmitting signal described in detecting step a.

As shown in figure 9, this method specifically includes the following steps:

Step 901: receiving end receives the M transmitting signal that transmitting terminal emits respectively from M transmitting antenna, obtains N number of reception signal, and carry out channel estimation according to signal is received, obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna.

Step 902: calculating H⁽¹⁾=[h₁₁] inverse matrix (H⁽¹⁾)^-1, i.e. Q⁽¹⁾=(H⁽¹⁾)^-1。

Wherein, H⁽¹⁾Be H-matrix preceding 1 row before 1 column element constitute submatrix.In the present embodiment, the initial submatrix for recursion is the submatrix of 1 row 1 column.However, it is to be understood that in practical applications, the submatrix of (k < M) can be arranged as the initial submatrix for being used for recursion down to arbitrary k row k using 2 rows 2 column.

Recursion variable m=2 is set.

Step 903: judging whether to have obtained the inverse matrix of H-matrix, that is, judge whether m is greater than M, if it is, going to step 907；Otherwise, step 904 is gone to.

Step 904: obtaining the H of this recursion^(m)Matrix；The H of this recursion^(m)Matrix can be in the H of upper primary recursion^(m-1)Increase a line on the basis of matrix and a column obtain, is embodied as $H^{\left(m\right)}=\left[\begin{array}{cc}{H}^{(m-1)}& v_{m-1}\\ y_{m-1}^H& {\mathrm{\&beta;}}_m\end{array}\right].$

Wherein, H^(m)Matrix is the submatrix of the preceding m row m column of H-matrix, specially

H^(m-1)Matrix is the submatrix of the preceding m-1 row m-1 column of H-matrix, specially

v_m-1For H-matrix m arrange preceding m-1 form vectors, specially $v_{m-1}=\left[\begin{array}{c}{h}_{1m}\\ {\mathrm{<figure-callout\; id="2m"\; label="h"\; filenames="H2008101874159E00091.png"\; state="\{\{state\}\}">h</figure-callout>}}_{2m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ h_{(m-1)m}\end{array}\right];$ β_mFor the element that the m row m of H-matrix is arranged, specially h_mm；And y_m-1 ^HFor the row vectors of preceding m-1 of m row compositions of H-matrix, specially $y_{m-1}^H=\left[\begin{array}{cccc}{h}_{m1}& {\mathrm{<figure-callout\; id="2"\; label="h\; m"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">h</figure-callout>}}_m& \&CenterDot;\&CenterDot;\&CenterDot;& h_{m(m-1)}\end{array}\right].$

Submatrix H for recursion^(m)It can be extracted from matrix H.

In practical calculating process, do not need specifically to obtain H^(m)Complete structure, and only need to obtain v_m-1、β_mAnd y_m-1 ^H, v_m-1、β_mAnd y_m-1 ^HIt can directly extract and obtain from matrix H.

Step 905: calculating H^(m)Inverse matrix (H^(m))^-1=Q^(m)。

First by Q_m-1、v_m-1、β_mAnd y_m-1 ^HCalculate μ_m-1=-Q_m-1v_m-1、 ${\mathrm{\&omega;}}_{m-1}=\frac{1}{{\mathrm{\&beta;}}_m-y_{m-1}^H{Q}_{m-1}{v}_{m-1}}$ With ${\mathrm{\&psi;}}_{m-1}^H=-y_{m-1}^H{Q}_{m-1},$ Again by μ_m-1、ψ_m-1 ^HAnd ω_m-1Pass through T_m-1=Q_m-1+ω_m-1μ_m-1ψ_m-1 ^HCalculate T_m-1, then by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}\\ {\mathrm{\&omega;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}\end{array}\right]$ Obtain Q_m。

And when specific implementation, it is above-mentioned by Q in order to reduce computation complexity_m-1Obtain Q_mRecurrence method can refine are as follows:

μ is calculated first_m-1=-Q_m-1v_m-1With ${\mathrm{\&psi;}}_{m-1}^H=-y_{m-1}^H{Q}_{m-1},$ Again by μ_m-1It calculates ${\mathrm{\&omega;}}_{m-1}=\frac{1}{{\mathrm{\&beta;}}_m+y_{m-1}^H{\mathrm{\&mu;}}_{m-1}},$ Finally by μ_m-1、ψ_m-1 ^HAnd ω_m-1Pass through $T_{m-1}=Q_{m-1}+{\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H$ Calculate T_m-1, then by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}\\ {\mathrm{\&omega;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}\end{array}\right]$ Obtain Q_m。

Alternatively, above-mentioned by Q_m-1Obtain Q_mRecurrence method can also be refined as another realize step, it may be assumed that calculate first ${\mathrm{\&psi;}}_{m-1}^H=-y_{m-1}^H{Q}_{m-1}$ And μ_m-1=-Q_m-1v_m-1, then by ψ_m-1 ^HIt calculates ${\mathrm{\&omega;}}_{m-1}=\frac{1}{{\mathrm{\&beta;}}_m+{\mathrm{\&psi;}}_{m-1}^H{v}_{m-1}},$ Finally by μ_m-1、ψ_m-1 ^HAnd ω_m-1Pass through $T_{m-1}=Q_{m-1}+{\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H$ Calculate T_m-1, then by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}\\ {\mathrm{\&omega;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}\end{array}\right]$ Obtain Q_m。

The value of step 906:m increases by 1, i.e. m=m+1 then goes to step 903.

In recursive process, recursion variable m can also increase integer, such as m=m+2 or m=m+3 greater than 1 etc..It means that if the inverse matrix for calculating 2 × 2 submatrixs can directly calculate the inverse matrix of 4 × 4 submatrixs, or can directly calculate the inverse matrix of 5 × 5 submatrixs then in recursive process next time in this recursive process.

Step 907: obtaining the inverse matrix H of matrix H^-1=Q, obtained Q meet Q=H^-1Relationship.

Step 908: calculating the estimated value of emission signal vector

WhereinIt can pass through $\hat{s}\left(k\right)=\mathrm{Qx}\left(k\right)$ It obtains.The estimated valueThe linear force zero estimated value of referred to as s (k).

Step 909: to the estimated value of obtained emission signal vectorItems quantified, obtain it is each transmitting signal testing result.

Using the method for the signal detection that the embodiment of the present invention six provides, reduce required calculation amount, and then reduce the time delay for using and reducing signal detection of the hardware resource of receiver.

A kind of method that the embodiment of the present invention seven gives signal detection, the at least two transmitting signal of detection in multiple-input, multiple-output mimo system, the transmitting signal is emitted respectively by each different transmitting antenna of transmitting terminal and reaches receiving end by a channel, this method comprises:

A. at least two receiving antennas of receiving end receive the transmitting signal, obtain at least two and receive signal；

B. receiving end carries out channel estimation according to signal is received, and obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna；

C. the LDU factoring matrix of a submatrix of H-matrix H is calculated using channel matrix H, then the LDU factoring matrix of a submatrix of calculated H and the newly-increased matrix entries obtained according to channel matrix H are utilized, recursion acquires the LDU factoring matrix of a submatrix comprising the H and a bigger submatrix of the H of a submatrix greater than the H, the newly-increased matrix entries are the newly-increased matrix entries of a submatrix comprising the H and a submatrix of a bigger submatrix inverse matrix H of the H of a submatrix greater than the H；

D. using the LDU factoring matrix of a bigger submatrix of the obtained H of step c, transmitting signal described in detecting step a.

As shown in Figure 10, this method specifically includes step:

Step 1001: receiving end receives the M transmitting signal that transmitting terminal emits respectively from M transmitting antenna, obtains N number of reception signal, and carry out channel estimation according to signal is received, obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna.

Step 1002: calculating H⁽¹⁾=[h₁₁] inverse matrix LDU decompose, i.e. L⁽¹⁾=1, U⁽¹⁾=1 and D⁽¹⁾=(H⁽¹⁾)^-1。

Wherein, H⁽¹⁾Be H-matrix preceding 1 row before 1 column element constitute submatrix.In the present embodiment, the initial submatrix for recursion is the submatrix of 1 row 1 column.However, it is to be understood that in practical applications, the submatrix of (k < M) can be arranged as the initial submatrix for being used for recursion down to arbitrary k row k using 2 rows 2 column.

Recursion variable m=2 is set.

Step 1003: judging whether that the LDU for having obtained the inverse matrix of H-matrix is decomposed, that is, judge whether m is greater than M, if it is, going to step 1007；Otherwise, step 1004 is gone to.

Step 1004: obtaining the H of this recursion^(m)Matrix；H^(m)Matrix can be in the H of upper primary recursion^(m-1)On the basis of increase a line, one column obtain, be embodied as $H^{\left(m\right)}=\left[\begin{array}{cc}{H}^{(m-1)}& v_{m-1}\\ y_{m-1}^H& {\mathrm{\&beta;}}_m\end{array}\right].$

Wherein, H^(m)Matrix is the submatrix of the preceding m row m column of H-matrix, speciallyH^(m-1)Matrix is the submatrix of the preceding m-1 row m-1 column of H-matrix, specially

v_m-1For H-matrix m arrange preceding m-1 form vectors, specially $v_{m-1}=\left[\begin{array}{c}{h}_{1m}\\ {\mathrm{<figure-callout\; id="2m"\; label="h"\; filenames="H2008101874159E00091.png"\; state="\{\{state\}\}">h</figure-callout>}}_{2m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ h_{(m-1)m}\end{array}\right];$ β_mFor the element that the m row m of H-matrix is arranged, specially h_mm；And y_m-1 ^HFor the row vectors of preceding m-1 of m row compositions of H-matrix, specially $y_{m-1}^H=\left[\begin{array}{cccc}{h}_{m1}& {\mathrm{<figure-callout\; id="2"\; label="h\; m"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">h</figure-callout>}}_m& \&CenterDot;\&CenterDot;\&CenterDot;& h_{m(m-1)}\end{array}\right].$

Submatrix H for recursion^(m)It can be extracted from matrix H.

In practical calculating process, do not need specifically to obtain H^(m)Complete structure, and only need to obtain v_m-1、β_mAnd y_m-1 ^H, v_m-1、β_mAnd y_m-1 ^HIt can directly extract and obtain from matrix H.

Step 1005: by H^(m-1)Inverse matrix (H^(m-1))^-1=Q^(m-1)LDU decompose Q_m-1=L_m-1D_m-1U_m-1That is L_m-1、D_m-1And U_m-1, recurrence calculation H^(m)Inverse matrix (H^(m))^-1=Q^(m)LDU decompose Q_m=L_mD_mU_m.Acquire $L_m=\left[\begin{array}{cc}{L}_{m-1}& a_{m-1}\\ 0_{m-1}^{\mathrm{<figure-callout\; id="1"\; label="H"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">H</figure-callout>}}& 1\end{array}\right],$ $D_m=\left[\begin{array}{cc}{D}_{m-1}& 0_{m-1}\\ 0_{m-1}^H& d_{\mathrm{mm}}\end{array}\right],$ $U_m=\left[\begin{array}{cc}{U}_{m-1}& 0_{m-1}\\ b_{m-1}^{\mathrm{<figure-callout\; id="1"\; label="H"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">H</figure-callout>}}& 1\end{array}\right],$ And only demand obtain it is therein $d_{\mathrm{mm}}=1/({\mathrm{\&beta;}}_m-y_{m-1}^H{L}_{m-1}{D}_{m-1}{U}_{m-1}{v}_{m-1}),$ a_m-1=-L_m-1D_m-1U_m-1v_m-1With $b_{m-1}^H=-y_{m-1}^H{L}_{m-1}{D}_{m-1}{U}_{m-1},$ L can be found out_m、D_mAnd U_m。

And when specific implementation, it is above-mentioned by L in order to reduce computation complexity_m-1、D_m-1And U_m-1Obtain L_m、D_mAnd U_mRecurrence method can refine are as follows:

Q is calculated first_m-1=L_m-1D_m-1U_m-1, then by Q_m-1It calculates $d_{\mathrm{mm}}=1/({\mathrm{\&beta;}}_m-y_{m-1}^H{Q}_{m-1}{v}_{m-1}),$ a_m-1=-Q_m-1v_m-1With $b_{m-1}^H=-y_{m-1}^H{Q}_{m-1};$

Or it is above-mentioned by L_m-1、D_m-1And U_m-1Obtain L_m、D_mAnd U_mRecurrence method can also be refined as another realize step, i.e., calculate first ${\stackrel{\&OverBar;}{y}}_{m-1}^H=y_{m-1}^H{L}_{m-1}$ And v_m-1=U_m-1v_m-1, then by y_m-1 ^HAnd v_m-1It calculates $d_{\mathrm{mm}}=1/({\mathrm{\&beta;}}_m-{\stackrel{\&OverBar;}{y}}_{m-1}^H{D}_{m-1}{\stackrel{\&OverBar;}{v}}_{m-1}),$ a_m-1=-L_m-1D_m-1v_m-1With $b_{m-1}^H=-{\stackrel{\&OverBar;}{y}}_{m-1}^H{D}_{m-1}{U}_{m-1},$ L can be found out_m、D_mAnd U_m。

The value of step 1006:m increases by 1, i.e. m=m+1 then goes to step 1003.

In recursive process, recursion variable m can also increase integer, such as m=m+2 or m=m+3 greater than 1 etc..It means that if the inverse matrix for calculating 2 × 2 submatrixs can directly calculate the inverse matrix of 4 × 4 submatrixs, or can directly calculate the inverse matrix of 5 × 5 submatrixs then in recursive process next time in this recursive process.

Step 1007: obtaining the inverse matrix H of matrix H^-1The LDU of=Q is decomposed, and obtained LDU meets Q=H^-1The relationship of=LDU.

Step 1008: calculating the estimated value of emission signal vector

It can pass through $\hat{s}\left(k\right)=\mathrm{LDUx}\left(k\right)$ It calculates and obtains.The estimated value

The linear force zero estimated value of referred to as s (k).

Step 1009: to the estimated value of obtained emission signal vector

Items quantified, obtain it is each transmitting signal testing result.Quantizing process is similar with the quantizing process in the step 306 of the embodiment of the present invention two, and this will not be repeated here.

Using the method for the signal detection that the embodiment of the present invention seven provides, reduce required calculation amount, and then reduce the time delay for using and reducing signal detection of the hardware resource of receiver.

More generally, the embodiment of the present invention six has also provided the method for seeking the inverse matrix of a general M × Metzler matrix, and the method that the embodiment of the present invention seven has also provided the LDU for the inverse matrix for seeking a general M × Metzler matrix to decompose.Computation complexity needed for these methods is lower than existing method.And unlike the method provided from the embodiment of the present invention four, the general M × Metzler matrix H being previously mentioned in the embodiment of the present invention six and embodiment seven does not need the associate matrix H for meeting the matrix H Yu the matrix^HThis equal condition.

The method that the inverse matrix for seeking a general M × Metzler matrix of the offer of the embodiment of the present invention six is described below, that is, the method for providing the inverse matrix of calculating matrix H specifically include as shown in figure 11:

Step 1101: calculating H⁽¹⁾=[h₁₁] inverse matrix (H⁽¹⁾)^-1, i.e. Q⁽¹⁾=(H⁽¹⁾)^-1。

Wherein, H⁽¹⁾Be H-matrix preceding 1 row before 1 column element constitute submatrix.In the present embodiment, the initial submatrix for recursion is the submatrix of 1 row 1 column.Certainly, in practical applications, the submatrix of (k < M) can be arranged as the initial submatrix for being used for recursion down to arbitrary k row k using 2 rows 2 column.

Recursion variable m=2 is set.

Step 1102: judging whether to have obtained the inverse matrix of H-matrix, that is, judge whether m is greater than M, if it is, going to step 1106；Otherwise, step 1103 is gone to.

Step 1103: obtaining the H of this recursion^(m)Matrix；H^(m)Matrix can pass through the H in upper primary recursion^(m-1)On the basis of increase a line, one column obtain, be embodied as $H^{\left(m\right)}=\left[\begin{array}{cc}{H}^{(m-1)}& v_{m-1}\\ y_{m-1}^H& {\mathrm{\&beta;}}_m\end{array}\right].$

Wherein, H^(m)Matrix is the submatrix of the preceding m row m column of H-matrix, specially

H^(m-1)Matrix is the submatrix of the preceding m-1 row m-1 column of H-matrix, specially

v_m-1For H-matrix m arrange preceding m-1 form vectors, specially $v_{m-1}=\left[\begin{array}{c}{h}_{1m}\\ {\mathrm{<figure-callout\; id="2m"\; label="h"\; filenames="H2008101874159E00091.png"\; state="\{\{state\}\}">h</figure-callout>}}_{2m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ h_{(m-1)m}\end{array}\right];$ β_mFor the element that the m row m of H-matrix is arranged, specially h_mm；And y_m-1 ^HFor the row vectors of preceding m-1 of m row compositions of H-matrix, specially $y_{m-1}^H=\left[\begin{array}{cccc}{h}_{m1}& {\mathrm{<figure-callout\; id="2"\; label="h\; m"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">h</figure-callout>}}_m& \&CenterDot;\&CenterDot;\&CenterDot;& h_{m(m-1)}\end{array}\right].$

Submatrix H for recursion^(m)It needs to extract from matrix H.

In practical calculating process, do not need specifically to obtain H^(m)Complete structure, and only need to obtain v_m-1、β_mAnd y_m-1 ^H, v_m-1、β_mAnd y_m-1 ^HIt can directly extract and obtain from matrix H.

Step 1104: calculating H^(m)Inverse matrix (H^(m))^-1=Q^(m)。

First by Q_m-1、v_m-1、β_mAnd y_m-1 ^HCalculate μ_m-1=-Q_m-1v_m-1、 ${\mathrm{\&omega;}}_{m-1}=\frac{1}{{\mathrm{\&beta;}}_m-y_{m-1}^H{Q}_{m-1}{v}_{m-1}}$ With ${\mathrm{\&psi;}}_{m-1}^H=-y_{m-1}^H{Q}_{m-1},$ Again by μ_m-1、ψ_m-1 ^HAnd ω_m-1Pass through $T_{m-1}=Q_{m-1}+{\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H$ Calculate T_m-1, then by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}\\ {\mathrm{\&omega;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}\end{array}\right]$ Obtain Q_m。

And when specific implementation, in order to reduce calculation amount, computation complexity is reduced, it is above-mentioned by Q_m-1Obtain Q_mRecurrence method can refine are as follows:

μ is calculated first_m-1=-Q_m-1v_m-1With ${\mathrm{\&psi;}}_{m-1}^H=-y_{m-1}^H{Q}_{m-1},$ Again by μ_m-1It calculates ${\mathrm{\&omega;}}_{m-1}=\frac{1}{{\mathrm{\&beta;}}_m+y_{m-1}^H{\mathrm{\&mu;}}_{m-1}},$ Finally by μ_m-1、ψ_m-1 ^HAnd ω_m-1Pass through $T_{m-1}=Q_{m-1}+{\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H$ Calculate T_m-1, then by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}\\ {\mathrm{\&omega;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}\end{array}\right]$ Obtain Q_m。

Alternatively, above-mentioned by Q_m-1Obtain Q_mRecurrence method can also be refined as another realize step, it may be assumed that calculate first ${\mathrm{\&psi;}}_{m-1}^H=-y_{m-1}^H{Q}_{m-1}$ And μ_m-1=-Q_m-1v_m-1, then by ψ_m-1 ^HIt calculates ${\mathrm{\&omega;}}_{m-1}=\frac{1}{{\mathrm{\&beta;}}_m+{\mathrm{\&psi;}}_{m-1}^H{v}_{m-1}},$ Finally by μ_m-1、ψ_m-1 ^HAnd ω_m-1Pass through $T_{m-1}=Q_{m-1}+{\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H$ Calculate T_m-1, then by $Q_m=\left[\begin{array}{cc}{T}_{m-1}& {\mathrm{\&omega;}}_{m-1}{\mathrm{\&mu;}}_{m-1}\\ {\mathrm{\&omega;}}_{m-1}{\mathrm{\&psi;}}_{m-1}^H& {\mathrm{\&omega;}}_{m-1}\end{array}\right]$ Obtain Q_m。

The value of step 1105:m increases by 1, i.e. m=m+1 then goes to step 1102.

In recursive process, recursion variable m can also increase integer, such as m=m+2 or m=m+3 greater than 1 etc..It means that if the inverse matrix for calculating 2 × 2 submatrixs can directly calculate the inverse matrix of 4 × 4 submatrixs, or can directly calculate the inverse matrix of 5 × 5 submatrixs then in recursive process next time in this recursive process.

Step 1106: obtaining the inverse matrix H of matrix H^-1=Q, obtained Q meet Q=H^-1Relationship.

The method that the LDU for the inverse matrix for seeking a general M × Metzler matrix that the embodiment of the present invention seven provides is decomposed, that is, the method for providing the LDU decomposition of the inverse matrix of calculating matrix H specifically include as shown in figure 12:

Step 1201: calculating H⁽¹⁾=[h₁₁] inverse matrix LDU decompose, i.e. L⁽¹⁾=1, U⁽¹⁾=1 and D⁽¹⁾=(H⁽¹⁾)^-1。

Wherein, H⁽¹⁾Be H-matrix preceding 1 row before 1 column element constitute submatrix.In the present embodiment, the initial submatrix for recursion is the submatrix of 1 row 1 column.Certainly, in practical applications, the submatrix of (k < M) can be arranged as the initial submatrix for being used for recursion down to arbitrary k row k using 2 rows 2 column.

Recursion variable m=2 is set.

Step 1202: judging whether that the LDU for having obtained the inverse matrix of H-matrix is decomposed, that is, judge whether m is greater than M, if it is, going to step 1206；Otherwise, step 1203 is gone to.

Step 1203: obtaining the H of this recursion^(m)Matrix；H^(m)Matrix can pass through the H in upper primary recursion^(m-1)On the basis of increase a line, one column obtain, be embodied as $H^{\left(m\right)}=\left[\begin{array}{cc}{H}^{(m-1)}& v_{m-1}\\ y_{m-1}^H& {\mathrm{\&beta;}}_m\end{array}\right]$

Wherein, H^(m)Matrix is the submatrix of the preceding m row m column of H-matrix, speciallyH^(m-1)Matrix is the submatrix of the preceding m-1 row m-1 column of H-matrix, specially

v_m-1For H-matrix m arrange preceding m-1 form vectors, specially $v_{m-1}=\left[\begin{array}{c}{h}_{1m}\\ {\mathrm{<figure-callout\; id="2m"\; label="h"\; filenames="H2008101874159E00091.png"\; state="\{\{state\}\}">h</figure-callout>}}_{2m}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ h_{(m-1)m}\end{array}\right];$ β_mFor the element that the m row m of H-matrix is arranged, specially h_mm；And y_m-1 ^HFor the row vectors of preceding m-1 of m row compositions of H-matrix, specially $y_{m-1}^H=\left[\begin{array}{cccc}{h}_{m1}& {\mathrm{<figure-callout\; id="2"\; label="h\; m"\; filenames="H2008101874159E00011.png,H2008101874159E00081.png"\; state="\{\{state\}\}">h</figure-callout>}}_m& \&CenterDot;\&CenterDot;\&CenterDot;& h_{m(m-1)}\end{array}\right].$

Submatrix H for recursion^(m)It needs to extract from matrix H.

In practical calculating process, do not need specifically to obtain H^(m)Complete structure, and only need to obtain v_m-1、β_mAnd y_m-1 ^H, v_m-1、β_mAnd y_m-1 ^HIt can directly extract and obtain from matrix H.

Step 1204: by H^(m-1)Inverse matrix (H^(m-1))^-1=Q^(m-1)LDU decompose Q_m-1=L_m-1D_m-1U_m-1That is L_m-1、D_m-1And U_m-1, recurrence calculation H^(m)Inverse matrix (H^(m))^-1=Q^(m)LDU decompose Q_m=L_mD_mU_m.Acquire $L_m=\left[\begin{array}{cc}{L}_{m-1}& a_{m-1}\\ 0_{m-1}^{\mathrm{<figure-callout\; id="1"\; label="H"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">H</figure-callout>}}& 1\end{array}\right],$ $D_m=\left[\begin{array}{cc}{D}_{m-1}& 0_{m-1}\\ 0_{m-1}^H& d_{\mathrm{mm}}\end{array}\right],$ $U_m=\left[\begin{array}{cc}{U}_{m-1}& 0_{m-1}\\ b_{m-1}^{\mathrm{<figure-callout\; id="1"\; label="H"\; filenames="H2008101874159E00011.png,H2008101874159E00021.png"\; state="\{\{state\}\}">H</figure-callout>}}& 1\end{array}\right],$ And only demand obtain it is therein $d_{\mathrm{mm}}=1/({\mathrm{\&beta;}}_m-y_{m-1}^H{L}_{m-1}{D}_{m-1}{U}_{m-1}{v}_{m-1}),$ a_m-1=-L_m-1D_m-1U_m-1v_m-1With $b_{m-1}^H=-y_{m-1}^H{L}_{m-1}{D}_{m-1}{U}_{m-1},$ L can be found out_m、D_mAnd U_m。

And when specific implementation, in order to reduce calculation amount, computation complexity is reduced, it is above-mentioned by L_m-1、D_m-1And U_m-1Obtain L_m、D_mAnd U_mRecurrence method can refine are as follows:

Q is calculated first_m-1=L_m-1D_m-1U_m-1, then by Q_m-1It calculates $d_{\mathrm{mm}}=1/({\mathrm{\&beta;}}_m-y_{m-1}^H{Q}_{m-1}{v}_{m-1}),$ a_m-1=-Q_m-1v_m-1With $b_{m-1}^H=-y_{m-1}^H{Q}_{m-1};$

Or it is above-mentioned by L_m-1、D_m-1And U_m-1Obtain L_m、D_mAnd U_mRecurrence method can also be refined as another realize step, i.e., calculate first ${\stackrel{\&OverBar;}{y}}_{m-1}^H=y_{m-1}^H{L}_{m-1}$ And v_m-1=U_m-1v_m-1, then by y_m-1 ^HAnd v_m-1It calculates $d_{\mathrm{mm}}=1/({\mathrm{\&beta;}}_m-{\stackrel{\&OverBar;}{y}}_{m-1}^H{D}_{m-1}{\stackrel{\&OverBar;}{v}}_{m-1}),$ a_m-1=-L_m-1D_m-1v_m-1With $b_{m-1}^H=-{\stackrel{\&OverBar;}{y}}_{m-1}^H{D}_{m-1}{U}_{m-1},$ L can be found out_m、D_mAnd U_m。

The value of step 1205:m increases by 1, i.e. m=m+1 then goes to step 1202.

In recursive process, recursion variable m can also increase integer, such as m=m+2 or m=m+3 greater than 1 etc..It means that if the inverse matrix for calculating 2 × 2 submatrixs can directly calculate the inverse matrix of 4 × 4 submatrixs, or can directly calculate the inverse matrix of 5 × 5 submatrixs then in recursive process next time in this recursive process.

Step 1206: obtaining the inverse matrix H of matrix H^-1The LDU of=Q is decomposed, and obtained LDU meets Q=H^-1The relationship of=LDU.

It should be understood that the method that the LDU for the inverse matrix for seeking a general M × Metzler matrix that the embodiments of the present invention seven provide is decomposed, can also be used to ask an inverse of a matrix matrix to get the inverse matrix H for arriving matrix H^-1After L, D and U matrix that the LDU of=Q is decomposed, then by matrix multiplication, L, D is multiplied with U, obtain an inverse of a matrix matrix Q=LDU.

Those of ordinary skill in the art will appreciate that: realizing all or part of the steps of above method embodiment, this can be accomplished by hardware associated with program instructions, foregoing routine can be stored in a computer readable storage medium, the program when being executed, executes step including the steps of the foregoing method embodiments；And storage medium above-mentioned includes: the various media that can store program code such as ROM, RAM, magnetic or disk.

The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, and all within the spirits and principles of the present invention, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention.

Claims (2)

1. a kind of method for detecting signal in multi-antenna digital wireless communication system, the at least two transmitting signal of detection in multiple-input, multiple-output mimo system, the transmitting signal is emitted respectively by each different transmitting antenna of transmitting terminal and reaches receiving end by a channel, it is characterised in that

A. at least two receiving antennas of receiving end receive the transmitting signal, obtain at least two and receive signal；

B. receiving end carries out channel estimation according to signal is received, and obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna；

C. a sub- inverse of a matrix matrix of matrix H is calculated using channel matrix H, then a sub- inverse of a matrix matrix of calculated matrix H and the newly-increased matrix entries obtained according to channel matrix H are utilized, recursion acquires the inverse matrix of a submatrix comprising the H and a bigger submatrix of the H of a submatrix greater than the H, and the newly-increased matrix entries are the newly-increased matrix entries of a submatrix comprising the H and a submatrix of a bigger submatrix inverse matrix H of the H of a submatrix greater than the H；

D. using the inverse matrix of a bigger submatrix of the obtained H of step c, transmitting signal described in detecting step a.

2. a kind of method for detecting signal in multi-antenna digital wireless communication system, the at least two transmitting signal of detection in multiple-input, multiple-output mimo system, the transmitting signal is emitted respectively by each different transmitting antenna of transmitting terminal and reaches receiving end by a channel, it is characterised in that

A. at least two receiving antennas of receiving end receive the transmitting signal, obtain at least two and receive signal；

B. receiving end carries out channel estimation according to signal is received, and obtains the channel matrix H being made of the channel coefficients between transmitting antenna and receiving antenna；

C. the LDU factoring matrix of a submatrix of matrix H is calculated using channel matrix H, then the LDU factoring matrix of a submatrix of calculated H and the newly-increased matrix entries obtained according to channel matrix H are utilized, recursion acquires the LDU factoring matrix of a submatrix comprising the H and a bigger submatrix of the H of a submatrix greater than the H, and the newly-increased matrix entries are the newly-increased matrix entries of a submatrix comprising the H and a submatrix of a bigger submatrix inverse matrix H of the H of a submatrix greater than the H；

D. using the LDU factoring matrix of a bigger submatrix of the obtained H of step c, transmitting signal described in detecting step a. 

Images