CN101834827B - Signal detection method and device in multiple-input multiple-output system - Google Patents

Abstract

The embodiment of the invention discloses signal detection method and device in a multiple-input multiple-output system. The signal detection method comprises the following steps of: converting a constellation point of a quadrature amplitude modulation (QAM) transmitting signal into a bit vector weight sum form to obtain a bit-level-expressed QAM transmitting signal; converting a channel matrix into a compound channel matrix according to the bit-level-expressed QAM transmitting signal; carrying out QR decomposition on the compound channel matrix to obtain an upper triangular matrix; constructing a bit-by-bit layered tree structure by using the upper triangular matrix; searching the bit-by-bit layered tree structure layer by layer based on a breadth-first M algorithm to obtain a signal candidate set; calculating a measure value of each candidate signal in the signal candidate set; and calculating the posterior information of the transmitting signal by using the measure value. The embodiment of the invention can reduce the realization difficulty of a signal detection process.

Description

Signal detecting method in a kind of multi-input multi-output system and device

Technical field

The application relates to communication and field of computer technology, particularly relates to signal detecting method and device in a kind of multi-input multi-output system.

Background technology

Along with development of wireless communication devices; Adopt the MIMO (Multiple-InputMultiple-Out-put of many bays; Multiple-input and multiple-output) technology has been owing to made full use of space resources, and can in limited bandwidth, improve the frequency efficiency of system greatly, therefore; The capacity potentiality that the MIMO technology provides single-antenna technology to provide for the user, and become an effective way of utilizing Spatial Dimension to improve power system capacity and reliability.At present, the MIMO technology has become one of hot spot technology among 3G or the 4G.

In mimo system, because the performance of receiving terminal finally can greatly influence transmission rate, bit error rate performance and the system complexity of whole transmission system, therefore, the improvement of receiving terminal performance also just becomes the focus that the researcher pays close attention to.At present, the researcher has proposed the sequential decoding method is applied in the signal detection process of mimo system.Wherein, the tree searching method of application breadth-first M algorithm can be searched for and obtain a signal candidate sequence, and the metric of signal calculated candidate sequence, utilizes these metrics to calculate the posterior information of the institute's corresponding bit that transmits, and accomplishes signal detection process.This tree searching method based on breadth-first M algorithm not only makes the complexity of calculating not change with the variation of signal to noise ratio and channel condition, but also makes transmission system have good bit error rate performance.

But; The inventor finds under study for action: the tree searching method of using breadth-first M algorithm is based on that other signal tree of symbol level searches for; Under constellation dimension condition with higher, the computation complexity of whole tree search procedure is high, has finally increased the realization difficulty of signal detection process.

Summary of the invention

In order to solve the problems of the technologies described above, the application embodiment provides signal detecting method and the device in a kind of multi-input multi-output system, to reduce the realization difficulty of signal detection process.

The application embodiment discloses following technical scheme:

Signal detecting method in a kind of multi-input multi-output system; Comprise: the form that the constellation point of quadrature amplitude modulation QAM transmission signal is converted into the bit vectors weighted sum; Obtain the QAM transmission signal that bit-level is represented; And, convert channel matrix into the compound channel matrix according to the QAM transmission signal that said bit-level is represented; Said compound channel matrix is carried out QR decompose, obtain a upper triangular matrix, utilize said upper triangular matrix to make up a tree structure by the bit layering; Based on breadth-first M algorithm the tree structure by the bit layering is successively searched for, obtained the signal candidate collection, calculate the metric of each candidate signal in the said signal candidate collection; Utilize said metric to calculate the posterior information that transmits.

Signal supervisory instrument in a kind of multi-input multi-output system; Comprise: converting unit; Be used for the constellation point of quadrature amplitude modulation QAM transmission signal is converted into the form of bit vectors weighted sum; Obtain the QAM transmission signal that bit-level is represented, and send signal, obtain the compound channel matrix according to the QAM that said bit-level is represented; Resolving cell is used for that said compound channel matrix is carried out QR and decomposes, and obtains a upper triangular matrix, utilizes said upper triangular matrix to make up a tree structure by the bit layering; Search unit is used for based on breadth-first M algorithm the tree structure by the bit layering successively being searched for, and obtains the signal candidate collection, calculates the metric of each candidate signal in the said signal candidate collection; Detecting unit utilizes said metric to calculate the posterior information that transmits.

Can find out that by the foregoing description bit-level tree searching and detecting method need can reduce the individual path of expansion search among the application at every turn, only need calculate the tolerance of 2M node after corresponding every node layer launches.In existing tree searching method based on symbol, in each stage of tree search, each is retained the path and all will expand to according to the size of constellation collection

Individual individual path after corresponding every node layer launches, needs to calculate these

The tolerance of node, and then make the complexity and the order of modulation M of whole tree search algorithm _rBe exponential relation, the application's implementation has reduced the whole algorithm complexity, thereby has reduced the realization difficulty of signal detection process.

Description of drawings

In order to be illustrated more clearly in the application embodiment or technical scheme of the prior art; To do to introduce simply to the accompanying drawing of required use in embodiment or the description of the Prior Art below; Obviously, the accompanying drawing in describing below only is some embodiment of the application, for those of ordinary skills; Under the prerequisite of not paying creative work property, can also obtain other accompanying drawing according to these accompanying drawings.

Fig. 1 is the flow chart of an embodiment of the signal detecting method in a kind of multi-input multi-output system of the application;

Fig. 2 is a kind of gray mappings sketch map of the application;

Fig. 3 is a kind of mapping naturally of the application sketch map;

It is the expansion search graph of 4 o'clock bit-level tree search algorithm that Fig. 4 retains the path for the application 16QAM constellation;

Fig. 5 is the structure chart of the embodiment of signal supervisory instrument in a kind of multi-input multi-output system of the application;

Fig. 6 is the structure chart of another embodiment of the signal supervisory instrument in a kind of multi-input multi-output system of the application.

Embodiment

For above-mentioned purpose, the feature and advantage that make the application can be more obviously understandable, the application embodiment is described in detail below in conjunction with accompanying drawing.

Embodiment one

See also Fig. 1, it is the flow chart of an embodiment of the signal detecting method in a kind of multi-input multi-output system of the application, and this method may further comprise the steps:

Step 101: the form that the constellation point of quadrature amplitude modulation QAM transmission signal is converted into the bit vectors weighted sum; Obtain the QAM transmission signal that bit-level is represented; And, convert channel matrix into the compound channel matrix based on the QAM transmission signal that said bit-level is represented;

Step 102: said compound channel matrix is carried out QR decompose, obtain a upper triangular matrix, utilize said upper triangular matrix to make up a tree structure by the bit layering;

Step 103: use breadth-first M algorithm the tree structure by the bit layering is successively searched for, obtain the signal candidate collection, calculate the metric of each candidate signal in the said signal candidate collection.

Step 104: utilize said metric to calculate the posterior information that transmits.

Be elaborated in the face of above-mentioned signal detection process down.Wherein, When the constellation point of quadrature amplitude modulation QAM being sent signal converts the form that bit vectors is represented into; Obtain the QAM transmission signal that bit vectors is represented; And, when converting channel matrix into the compound channel matrix, can carry out according to following mode according to the QAM transmission signal that said bit vectors is represented.

Suppose to use N at one _tTransmit antennas and N _rIn the mimo system model of root reception antenna, sending symbol is the quadrature amplitude modulation of square constellation, and the size of constellation collection does M wherein _cSend the bit number of symbol under complex field for each, then, the discrete-time system model of mimo system under complex field is:

y ^c＝H ^cs ^c+n ^c (1)

With the system model equivalence under the complex field of formula (1) is that a system model under real number field is:

y＝Hs+n (2)

Can draw, under real number field, the vectorial dimension that receives signal y is N _R* 1, the vectorial dimension of sending signal s is N _T* 1, the vectorial dimension of channel matrix H is N _T* N _T, the vectorial dimension of real Gaussian noise n is N _R* 1, average does Variance matrix does

Wherein, N _RBe the equivalent received number of antennas under the real number field, N _TBe the equivalent transmitting antenna number under the real number field, and N _R=2N _r, N _T=2N _t

When mimo channel is the Rayleigh channel of flat fading, every frame N _TM _rLong bit symbol is:

$x={\left[\begin{array}{cccccccccc}{x}_{\mathrm{1,1}}& \·\·\·& x_{1,M_r}& x_{\mathrm{2,1}}& \·\·\·& x_{2,M_r}& \·\·\·& x_{N_T,1}& \·\·\·& x_{N_T,M_r}\end{array}\right]}^T---\left(3\right)$

Wherein, M _rFor each sends the bit number of symbol under real number field, and M _r=M _c/ 2.Send symbol s through being mapped as one behind the serial to parallel conversion.

When system uses gray mappings as shown in Figure 2, the transmission symbol s on the k transmit antennas then _kWith its corresponding mapped bits symbol x _{K, j}, j=1 ..., M _rBetween relation can represent by following formula:

$s_k=\underset{i=1}{\overset{M_r}{\mathrm{\Σ}}}{w}_i\underset{j=1}{\overset{i}{\mathrm{\Π}}}{x}_{k,j}---\left(7\right)$

Fig. 2 has provided an example of 8-PAM gray mappings constellation, has introduced a middle parameter u here _{K, i}, u _{K, i}∈+1, and-1}, order,

$u_{k,i}=\underset{j=1}{\overset{i}{\mathrm{\Σ}}}{x}_{k,j},i=\mathrm{1,2},\·\·\·,M_r---\left(8\right)$

When using u _kRepresent u _{K, i}, i=1 ..., M _rThe vector of being formed is represented corresponding to u with w _{K, i}Weight factor w _{I, i}=1 ..., M _rFormed vectorial the time, then,

$u_k={\left[\begin{array}{cccc}{u}_{k,1}& u_{k,2}& \·\·\·& u_{k,M_r}\end{array}\right]}^T,$ $w={\left[\begin{array}{cccc}{w}_1& w_2& \·\·\·& w_{M_r}\end{array}\right]}^T---\left(9\right)$

Therefore, s _kVector representation be:

s _k＝w ^Tu _k (10)

For M-QAM,

Weight factor:

$w=\sqrt{\frac{3}{2(M-1)}}{\left[\begin{array}{cccccccc}\sqrt{M}{2}^{-1},& \sqrt{M}{2}^{-2},& \·\·\·,& 1,& \sqrt{M}{2}^{-1}j,& \sqrt{M}{2}^{-2}j,& \·\·\·& j\end{array}\right]}^T---\left(11\right)$

Wherein, J represents imaginary symbols, and promptly

M is a number of constellation points.For example; When using the 16QAM constellation; M=16,

From u _{K, i}Definition can draw x _{K, i}With u _{K, i}And u _{K, i-1}Between following relation arranged:

x _k，i＝u _k，iu _k，i-1，i＝2，…，M _r (12)

When i=1, x _{K, i}=u _{K, i}In formula (12) substitution formula (7), can obtain the relation between vectorial s and the u:

$s=(I_{N_T}\⊗w^T)u---\left(13\right)$

Wherein,

expression Kronecker direct product.

Formula (13) is brought in the formula (2),

then arranged

The constellation point that quadrature amplitude modulation QAM is sent signal converts the form of bit vectors weighted sum into, obtains the QAM that bit-level representes and sends signal u, and according to the relation that QAM sends signal s and bit vectors u, define the compound channel matrix:

$A=H(I_{N_T}\⊗w^T)$

$=\left[\begin{array}{cccc}{\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="HSA00000066275800012.png,HSA00000066275800021.png"\; state="\{\{state\}\}">h</figure-callout>}}_1\&CircleTimes;w^T& h_2\&CircleTimes;w^T& \&CenterDot;\&CenterDot;\&CenterDot;& h_{N_T}\&CircleTimes;w^T\end{array}\right]---\left(14\right)$

Wherein, h _iRepresent the i row of channel matrix H.Definition N _c=N _TM _r, the dimension of compound channel matrix A is N _R* N _cThe equivalent channel model just can be written as so:

y＝Au+n (15)

In addition, when system uses as shown in Figure 3 shining upon naturally, the symbol s on the k transmit antennas then _kWith its corresponding mapped bits x _{K, j}, j=1 ..., M _rBetween relation can represent by following formula:

$s_k=\underset{i=1}{\overset{M_c}{\mathrm{\&Sigma;}}}{w}_i{x}_{k,i}---\left(16\right)$

s _kVector representation be: s _k=w ^Tx _k(17)

Wherein, $x_k={\left[\begin{array}{cccc}{x}_{k,1}& x_{k,2}& \&CenterDot;\&CenterDot;\&CenterDot;& x_{k,M_r}\end{array}\right]}^T.$

When using the square qam constellation, the weight factor w in the formula (17) calculates according to the mode of formula (11).So, be similar to gray mappings, definition compound channel matrix is:

$A=H(I_{N_T}\&CircleTimes;w^T)$

$=\left[\begin{array}{cccc}{\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="HSA00000066275800012.png,HSA00000066275800021.png"\; state="\{\{state\}\}">h</figure-callout>}}_1\&CircleTimes;w^T& h_2\&CircleTimes;w^T& \&CenterDot;\&CenterDot;\&CenterDot;& h_{N_T}\&CircleTimes;w^T\end{array}\right]$

The dimension of compound channel matrix A is N _R* N _cThe channel model of equivalence just can be written as so:

y＝Ax+n (18)

Wherein, $x={\left[\begin{array}{cccc}{x}_1^T& x_2^T& \&CenterDot;\&CenterDot;\&CenterDot;& x_{N_T}^T\end{array}\right]}^T$

Decompose when said compound channel matrix being utilized the MMSE criterion carry out QR, obtain a upper triangular matrix, when utilizing one of said upper triangular matrix structure to pursue the tree structure of bit layering, can carry out according to following mode.

At first, in order to obtain the non-singular matrix of an equivalence, based on the MMSE criterion, definition expansion compound channel matrix is:

$\underset{\&OverBar;}{A}\stackrel{\mathrm{\&Delta;}}{=}\left[\begin{array}{c}A\\ {\mathrm{\&sigma;}}_n{I}_{N_c}\end{array}\right]---\left(19\right)$

Secondly, to expansion compound channel matrix ACarrying out the QR decomposition obtains A=QR, here Its column vector is mutually orthogonal, and

It is a upper triangular matrix.

Here need to prove that in the signal detection process in multi-input multi-output system, tolerance is calculated according to following formula:

$\mathrm{\&mu;}\left(s\right)=-\frac{1}{2{\mathrm{\&sigma;}}_n^2}{\left|\right|y-\mathrm{Hs}\left|\right|}^2+\frac{1}{2}{x}^T{L}_A\left(x\right)---\left(20\right)$

Wherein, L _A(x) expression and the corresponding prior information of bit vectors x.

At last, in order better to be detected effect, before QR decomposed, we were to matrix AEach column weight newly sort.

For gray mappings, with in the formula (20) || y-Hs|| ²With

Associate, the metric when obtaining gray mappings,

${\left|\right|y-\mathrm{Hs}\left|\right|}^2={\left|\right|y-\mathrm{Au}\left|\right|}^2$

$=y^{H}y-y^H\mathrm{Au}-u^H{A}^{H}y+u^H{A}^H\mathrm{Au}$

$=y^{H}y-{\hat{u}}^H{R}^H\mathrm{Ru}-u^H{R}^{H}R\hat{u}+u^H(R^{H}R-{\mathrm{\&sigma;}}_n^2{I}_{N_c})u$

$={\hat{u}}^H{R}^{H}R\hat{u}-{\hat{u}}^H{R}^H\mathrm{Ru}-u^H{R}^{H}R\hat{u}+u^H{R}^H\mathrm{Ru}-{\mathrm{\&sigma;}}_n^2{u}^{H}u-{\hat{u}}^H{R}^{H}R\hat{u}+y^{H}y$

$={\left|\right|R(u-\hat{u})\left|\right|}^2-{\mathrm{\&sigma;}}_n^2{u}^{T}u-{\hat{u}}^H{R}^{H}R\hat{u}+y^{H}y$

$={\left|\right|R(u-\hat{u})\left|\right|}^2-{\mathrm{\&sigma;}}_n^2{u}^{T}u-y^H(A{(A^{H}A+{\mathrm{\&sigma;}}_n^2{I}_{N_c})}^{-1}{A}^H-I_{N_R})y---\left(21\right)$

Wherein, $\hat{u}={(A^{H}A+{\mathrm{\&sigma;}}_n^2{I}_{N_c})}^{-1}{A}^{H}y.$

Order $C={\mathrm{\&sigma;}}_n^2{u}^{T}u+y^H(A{(A^{H}A+{\mathrm{\&sigma;}}_n^2{I}_{N_c})}^{-1}{A}^H-I_{N_R})y---\left(22\right)$

Because the symbol among the intermediate vector u all is a bit symbol, satisfy | u _{K, i}| ²=1, obviously this moment C size and intermediate variable u choose that it doesn't matter, therefore, can it be omitted from metric calculation.Tolerance just can be expressed as again so:

$\mathrm{\&mu;}\left(s\right)=-\frac{1}{2{\mathrm{\&sigma;}}_n^2}{\left|\right|R(u-\hat{u})\left|\right|}^2+\frac{1}{2}{x}^T{L}_A\left(x_{n,k}\right)$

$=\underset{i=1}{\overset{N_c}{\mathrm{\&Sigma;}}}(-\frac{1}{2{\mathrm{\&sigma;}}_n^2}{\left|\underset{j=1}{\overset{i}{\mathrm{\&Sigma;}}}{r}_{i,j}(u_j-{\hat{u}}_j)\right|}^2+\frac{1}{2}{x}_i{L}_A\left(x_i\right))---\left(23\right)$

In addition, for mapping naturally, with in the formula (20) || y-Hs|| ²With

Associate, the metric when obtaining the nature mapping,

$\mathrm{\&mu;}\left(s\right)=-\frac{1}{2{\mathrm{\&sigma;}}_n^2}{\left|\right|R(x-\hat{x})\left|\right|}^2+\frac{1}{2}{x}^T{L}_A\left(x\right)$

$=\underset{i=1}{\overset{N_c}{\mathrm{\&Sigma;}}}(-\frac{1}{2{\mathrm{\&sigma;}}_n^2}{\left|\underset{j=1}{\overset{i}{\mathrm{\&Sigma;}}}{r}_{i,j}(x_j-{\hat{x}}_j)\right|}^2+\frac{1}{2}{x}_i{L}_A\left(x_i\right))---\left(24\right)$

Wherein, $\hat{x}={(A^{H}A+{\mathrm{\&sigma;}}_n^2{I}_{N_c})}^{-1}{A}^{H}y.$

Above-mentioned formula (23) and formula (24) are respectively gray mappings and shine upon the mathematic(al) representation of signal metric value down naturally; Promptly; Based on breadth-first M algorithm the tree structure by the bit layering is successively searched for; Can obtain the signal candidate collection, and calculate the metric process mathematical model of each candidate signal in the said signal candidate collection.

Wherein, Based on breadth-first M algorithm the tree structure by the bit layering is successively searched for; Obtain the signal candidate collection; The metric that calculates said signal candidate collection comprises: in the current layer of said tree structure by the bit layering, calculate the metric by 2M node on M retention 2M the individual path that node launched of last layer; M maximum node of search metric value from a said 2M node, M the node that said metric is maximum is as the retention node of current layer; Search successively successively selects M to retain node as the signal candidate collection from 2M the node of last one deck, calculates the metric of said signal candidate collection.

For example; As shown in Figure 4, at the k layer of the tree structure that pursues the bit layering, calculate 4 metrics of retaining 8 nodes on 8 individual paths that node launched by the k-1 layer; Relatively the metric of 8 nodes is big or small, and therefrom 4 nodes of selectance value maximum are as the retention node of k layer.Search successively successively when calculating last one deck, is selected 4 and is retained node from 8 nodes of last one deck, and retains node as the signal candidate collection, 4 metrics of retaining node in the signal calculated candidate collection with these 4.

In gray mappings, the update calculation of above-mentioned metric, can represent according to following mathematic(al) representation:

${\mathrm{\&mu;}}_1=-\frac{1}{2{\mathrm{\&sigma;}}_n^2}{\left|{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HSA00000066275800012.png,HSA00000066275800021.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,1}}(u_1-{\hat{u}}_1)\right|}^2+\frac{1}{2}{x}_1{L}_A\left(x_1\right)$

${\mathrm{\&mu;}}_i={\mathrm{\&mu;}}_{i-1}-\frac{1}{2{\mathrm{\&sigma;}}_n^2}{|r_{i,i}(u_i-{\hat{u}}_i)+\underset{j=1}{\overset{i-1}{\mathrm{\&Sigma;}}}{r}_{i,j}(u_j-{\hat{u}}_j)|}^2+\frac{1}{2}{x}_i{L}_A\left(x_i\right),2\&le;i\&le;N_c---\left(25\right)$

$\mathrm{\&mu;}\left(s\right)={\mathrm{\&mu;}}_{N_c}$

See tolerance layering calculation process from formula (25), and after each depth amount is calculated completion, measure ordering, keep M maximum node of tolerance, as the reservation node that launches next time.

In shining upon naturally, the update calculation of above-mentioned metric, can represent according to following mathematic(al) representation:

${\mathrm{\&mu;}}_1=-\frac{1}{2{\mathrm{\&sigma;}}_n^2}{\left|{\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HSA00000066275800012.png,HSA00000066275800021.png"\; state="\{\{state\}\}">r</figure-callout>}}_{\mathrm{1,1}}\left({x_1-\hat{x}}_1\right)\right|}^2+\frac{1}{2}{x}_1{L}_A\left(x_1\right)$

${\mathrm{\&mu;}}_i={\mathrm{\&mu;}}_{i-1}-\frac{1}{2{\mathrm{\&sigma;}}_n^2}{|r_{i,i}(x_i-{\hat{x}}_i)+\underset{j=1}{\overset{i-1}{\mathrm{\&Sigma;}}}{r}_{i,j}(x_j-{\hat{x}}_j)|}^2+\frac{1}{2}{x}_i{L}_A\left(x_i\right),2\&le;i\&le;N_c---\left(26\right)$

$\mathrm{\&mu;}\left(s\right)={\mathrm{\&mu;}}_{N_c}$

A kind ofly preferred embodiment be; Based on breadth-first M algorithm the tree structure by the bit layering is successively searched for; Obtain the signal candidate collection; The metric that calculates said signal candidate collection comprises: by in the tree structure of bit layering, obtain the prior information of all layers corresponding bit said from decoder; Prior information is not searched for expansion greater than the node institute respective layer of first threshold; And extend in the retention path that the symbol that uses the current layer prior information is retained node with last layer; The node that prior information is not more than first threshold launches; According to the size of metric, M node of search metric value maximum is as the retention node of current layer; Search successively successively selects M to retain node as the signal candidate collection from last one deck, calculates the metric of said signal candidate collection.

For example; By in the tree structure of bit layering, obtain the prior information of all layers corresponding bit from decoder, the prior information of all layers corresponding bit compares with first threshold respectively; For example; The prior information of K layer corresponding bit then use the symbol of k layer prior information to extend in 4 retention paths of retaining node of K-1 layer, and the prior information of K+1 layer corresponding bit is not more than first threshold greater than first threshold; Then 4 with the K layer retain the node expansion; Calculating is by the metric of 8 nodes on 48 individual paths that node launched, and relatively the metric of 8 nodes is big or small, and therefrom 4 nodes of selectance value maximum are as the retention node of k+1 layer.Search successively successively when searching last one deck, selects 4 to retain nodes from last one deck, and retains nodes as the signal candidate collection with these 4, calculates the metric of said signal candidate collection.

Another preferred embodiment is; Based on breadth-first M algorithm the tree structure by the bit layering is successively searched for; Obtain the signal candidate collection; The metric that calculates said signal candidate collection comprises: by in the tree structure of bit layering, obtain the prior information of all layers corresponding bit said from decoder; With prior information greater than the prior information weighted value of the node of second threshold value as metric; Retain M maximum node of search metric value 2M the node that node launches from M of the upper strata, with M node of said metric maximum retention node as current layer; Search successively successively selects M to retain node as the signal candidate collection from last one deck, calculates the metric of said signal candidate collection.

For example; In pursuing the tree structure of bit layering; Obtain the prior information of all layers corresponding bit from decoder; The prior information of all layers corresponding bit is compared with second threshold value respectively; With prior information greater than the prior information weighted value of the node of second threshold value as metric; As, for gray mappings,

in the formula (19) is the prior information weighted value; For mapping naturally,

in the formula (20) is the prior information weighted value.As, when the prior information of k layer corresponding bit greater than second threshold value, then will be by 4 in the k-1 layer prior information weighted values of retaining 8 nodes that node launched as metric, therefrom maximum 4 nodes of selectance value are as the retention node of k layer.。Search successively successively when searching last one deck, selects 4 to retain nodes from last one deck, and retains nodes as the signal candidate collection with these 4, calculates the metric of said signal candidate collection.

Certainly, also can combine two kinds of optimal ways to accomplish.As; When pursuing in the tree structure of bit layering; Obtain the prior information of all layers corresponding bit from decoder after, the prior information with all layers corresponding bit compares with first threshold respectively earlier, and prior information is not searched for expansion greater than the node institute respective layer of first threshold; As; When the prior information of k layer corresponding bit during greater than first threshold, use the symbol of k layer prior information to extend in 4 retention paths of retaining nodes of k-1 layer, from rest layers with prior information greater than the prior information weighted value of the node of second threshold value as metric.As, when the prior information of k layer corresponding bit greater than second threshold value, then will be by 4 in the k-1 layer prior information weighted values of retaining 8 nodes that node launched as metric, therefrom maximum 4 nodes of selectance value are as the retention node of k layer.And be not more than the node of second threshold value for prior information, and will be according to the size of metric, M maximum node of search metric value from 2M the node that launches is with M node of the metric maximum retention node as current layer.Search successively successively when searching last one deck, selects 4 to retain nodes from last one deck, and retains nodes as the signal candidate collection with these 4, calculates the metric of said signal candidate collection.

Need to prove; When using gray mappings; Can obtain from the definition of tolerance, if

we only need the calculating tolerance relevant with prior information to get final product so.Following relation is arranged between the tolerance:

$\frac{1}{{\mathrm{\&sigma;}}_n^2}{\left|\underset{j=1}{\overset{i}{\mathrm{\&Sigma;}}}{r}_{i,j}(u_j-{\hat{u}}_j)\right|}^2\&le;\frac{1}{{\mathrm{\&sigma;}}_n^2}{\left|\right|R(u-\hat{u})\left|\right|}^2=\frac{1}{{\mathrm{\&sigma;}}_n^2}({\left|\right|y-\mathrm{Hs}\left|\right|}^2+C)---\left(27\right)$

|| y-Hs|| ²The expectation of+C can be approximated to be σ _n ²(N _R+ N _c).When | L _A(x _i) |＞＞T ₁, T ₁>=N _R+ N _cThe time, wherein, T ₁Be second threshold value, just can only calculate the tolerance relevant with prior information, this moment, the tolerance relevant with the i layer signal can be written as:

${\mathrm{\&mu;}}_I=\frac{1}{2}{x}_i{L}_A\left(x_i\right)---\left(28\right)$

Along with the increase of signal to noise ratio, the prior information absolute value that obtains from channel decoder can increase.When the absolute value of the prior information that obtains from decoder surpasses one when setting thresholding (first threshold), can think that the prior information about these bits is reliable.We just no longer calculate its tolerance to these bits so, and this moment, the degree of depth of search tree just can reduce greatly.Therefore the computation complexity of algorithm can reduce with the increase of signal to noise ratio.

Above-mentioned principle is applicable to the nature mapping too.

Also need to prove, and the span of first threshold is 3～5 times second threshold value.

Can find out that by the foregoing description bit-level tree searching and detecting method need can reduce the individual path of expansion search among the application at every turn, only need calculate the tolerance of 2M node after corresponding every node layer launches.In existing tree searching method based on symbol, in each stage of tree search, each is retained the path and all will expand to according to the size of constellation collection Individual individual path after corresponding every node layer launches, needs to calculate these

The tolerance of node, and then make the complexity and the order of modulation M of whole tree search algorithm _rBe exponential relation, the application's implementation has reduced the whole algorithm complexity, thereby has reduced the realization difficulty of signal detection process.

Embodiment three

Corresponding with the signal detecting method in above-mentioned a kind of multi-input multi-output system, the application embodiment also provides the signal supervisory instrument in a kind of multi-input multi-output system.See also Fig. 5, it is the structure chart of the embodiment of signal supervisory instrument in the application's multi-input multi-output system, and this device comprises converting unit 501, resolving cell 502, search unit 503 and detecting unit 504.Operation principle below in conjunction with this device is further introduced its internal structure and annexation.

Converting unit 501; Be used for the constellation point of quadrature amplitude modulation QAM transmission signal is converted into the form of bit vectors weighted sum; Obtain the QAM transmission signal that bit-level is represented, and send signal, convert channel matrix into the compound channel matrix according to the QAM that said bit-level is represented;

Resolving cell 502 is used for that said compound channel matrix is carried out QR and decomposes, and obtains a upper triangular matrix, utilizes said upper triangular matrix to make up a tree structure by the bit layering;

Search unit 503 is used for based on breadth-first M algorithm the tree structure by the bit layering successively being searched for, and obtains the signal candidate collection, calculates the metric of each candidate signal in the said signal candidate collection;

Detecting unit 504 is used to utilize said metric to calculate the posterior information that transmits.

See also Fig. 6, it is the structure chart of another embodiment of the signal supervisory instrument in a kind of multi-input multi-output system of the application.Wherein, search unit 503 further comprises: computation subunit 5031, first search subelement 5032 and the chooser unit 5033,

Computation subunit 5031 is used for the current layer at said tree structure by the bit layering, calculates the metric by 2M node on M retention 2M the individual path that node launched of last layer;

The first search subelement 5032 is used for from M maximum node of said 2M node search metric value, and M the node that said metric is maximum is as the retention node of current layer;

Chooser unit 5033 is used for successively searching for successively, from 2M the node of last one deck, selects M to retain node as the signal candidate collection, calculates the metric of said signal candidate collection.

Preferably, aforementioned calculation subelement 5031 can be replaced by and obtain subelement, and the first search subelement 5032 can be replaced by the second search subelement, and then, search unit 503 comprises:

Obtain subelement, be used for obtaining the prior information of all layers corresponding bit from decoder at said tree structure by the bit layering;

The second search subelement; Be used for prior information is not searched for expansion greater than the node institute respective layer of first threshold; And extend in the retention path that the symbol that uses the current layer prior information is retained node with preceding one deck; The node that prior information is not more than first threshold launches, and according to the size of metric, M node of search metric value maximum is as the retention node of current layer;

The chooser unit is used for successively searching for successively, from last one deck, selects M to retain node as the signal candidate collection, calculates the metric of said signal candidate collection.

Preferably, aforementioned calculation subelement 5031 can be replaced by and obtain subelement, and the first search subelement 5032 can be replaced by the 3rd search subelement, and then, search unit 503 comprises:

Obtain subelement, be used for obtaining the prior information of all layers corresponding bit from decoder at said tree structure by the bit layering;

The 3rd search subelement; Be used for prior information greater than the prior information weighted value of the node of second threshold value as metric; Retain M maximum node of selectance value said 2M the node that node launches from M of the upper strata, with M node of said metric maximum retention node as current layer;

The chooser unit is used for successively searching for successively, from last one deck, selects M to retain node as the signal candidate collection, calculates the metric of said signal candidate collection.

The mapping mode that said quadrature amplitude modulation QAM sends the constellation point of signal comprises gray mappings or mapping naturally.

Can find out that by the foregoing description bit-level tree searching and detecting method need can reduce the individual path of expansion search among the application at every turn, only need calculate the tolerance of 2M node after corresponding every node layer launches.In existing tree searching method based on symbol, in each stage of tree search, each is retained the path and all will expand to according to the size of constellation collection

Individual individual path after corresponding every node layer launches, needs to calculate these

The tolerance of node, and then make the complexity and the order of modulation M of whole tree search algorithm _rBe exponential relation, the application's implementation has reduced the whole algorithm complexity, thereby has reduced the realization difficulty of signal detection process.

Need to prove; One of ordinary skill in the art will appreciate that all or part of flow process that realizes in the foregoing description method; Be to instruct relevant hardware to accomplish through computer program; Described program can be stored in the computer read/write memory medium, and this program can comprise the flow process like the embodiment of above-mentioned each side method when carrying out.Wherein, described storage medium can be magnetic disc, CD, read-only storage memory body (Read-Only Memory, ROM) or at random store memory body (Random AccessMemory, RAM) etc.

More than signal detecting method and device in a kind of multi-input multi-output system that the application provided are described in detail; Used specific embodiment herein the application's principle and embodiment are set forth, the explanation of above embodiment just is used to help to understand the application's method and core concept thereof; Simultaneously, for one of ordinary skill in the art, according to the application's thought, the part that all can change in specific embodiments and applications, in sum, this description should not be construed as the restriction to the application.

Claims (10)

1. the signal detecting method in the multi-input multi-output system is characterized in that, comprising:

The constellation point that quadrature amplitude modulation QAM is sent signal converts the form of bit vectors weighted sum into, obtains the QAM that bit-level representes and sends signal, and send signal according to the QAM that said bit-level is represented, converts channel matrix into the compound channel matrix;

Said compound channel matrix is carried out QR decompose, obtain a upper triangular matrix, utilize said upper triangular matrix to make up a tree structure by the bit layering;

Based on breadth-first M algorithm the tree structure by the bit layering is successively searched for, obtained the signal candidate collection, calculate the metric of each candidate signal in the said signal candidate collection;

Utilize said metric to calculate the posterior information that transmits.

2. method according to claim 1 is characterized in that, said based on breadth-first M algorithm to successively searching for by the tree structure of bit layering, obtain the signal candidate collection, the metric that calculates said signal candidate collection comprises:

In the current layer of said tree structure by the bit layering, calculate metric by 2M node on M retention 2M the individual path that node launched of last layer, said M is a number of constellation points;

M maximum node of search metric value from a said 2M node, M the node that said metric is maximum is as the retention node of current layer;

Search successively successively selects M to retain node as the signal candidate collection from 2M the node of last one deck, calculates the metric of said signal candidate collection.

3. method according to claim 1 is characterized in that, said based on breadth-first M algorithm to successively searching for by the tree structure of bit layering, obtain the signal candidate collection, the metric that calculates said signal candidate collection comprises:

By in the tree structure of bit layering, obtain the prior information of all layers corresponding bit said from decoder;

Prior information is not searched for expansion greater than the node institute respective layer of first threshold; And extend in the retention path that the symbol that uses the current layer prior information is retained node with preceding one deck; The node that prior information is not more than first threshold launches; According to the size of metric, M node of search metric value maximum is as the retention node of current layer, and said M is a number of constellation points;

Search successively successively selects M to retain node as the signal candidate collection from last one deck, calculates the metric of said signal candidate collection.

4. method according to claim 1 is characterized in that, said based on breadth-first M algorithm to successively searching for by the tree structure of bit layering, obtain the signal candidate collection, the metric that calculates said signal candidate collection comprises:

By in the tree structure of bit layering, obtain the prior information of all layers corresponding bit said from decoder;

With prior information greater than the prior information weighted value of the bit node of second threshold value as metric; From M of upper strata M node retaining search metric value maximum 2M the node that node launches; M the node that said metric is maximum is as the retention node of current layer, and said M is a number of constellation points;

Search successively successively selects M to retain node as the signal candidate collection from last one deck, calculates the metric of said signal candidate collection.

5. according to any described method of claim 1-4, it is characterized in that the mapping mode that said quadrature amplitude modulation QAM sends the constellation point of signal comprises gray mappings or mapping naturally.

6. the signal supervisory instrument in the multi-input multi-output system is characterized in that, comprising:

Converting unit is used for the constellation point that quadrature amplitude modulation QAM sends signal is converted into the form of bit vectors weighted sum, obtains the QAM that bit-level representes and sends signal, and send signal according to the QAM that said bit-level is represented, obtains the compound channel matrix;

Resolving cell is used for that said compound channel matrix is carried out QR and decomposes, and obtains a upper triangular matrix, utilizes said upper triangular matrix to make up a tree structure by the bit layering;

Search unit is used for based on breadth-first M algorithm the tree structure by the bit layering successively being searched for, and obtains the signal candidate collection, calculates the metric of each candidate signal in the said signal candidate collection;

Detecting unit utilizes said metric to calculate the posterior information that transmits.

7. device according to claim 6 is characterized in that, said search unit comprises:

Computation subunit is used for the current layer at said tree structure by the bit layering, calculates the metric by 2M node on M retention 2M the individual path that node launched of last layer, and said M is a number of constellation points;

The first search subelement is used for from M maximum node of said 2M node search metric value, and M the node that said metric is maximum is as the retention node of current layer;

The chooser unit is used for successively searching for successively, from 2M the node of last one deck, selects M to retain node as the signal candidate collection, calculates the metric of said signal candidate collection.

8. device according to claim 6 is characterized in that, said search unit comprises:

Obtain subelement, be used for obtaining the prior information of all layers corresponding bit from decoder at said tree structure by the bit layering;

The second search subelement; Be used for prior information is not launched search greater than the node institute respective layer of first threshold; And using the symbol of current layer prior information directly to be extended in the retention path of preceding one deck retention node, the node that prior information is not more than first threshold launches, according to the size of metric; M node of search metric value maximum is as the retention node of current layer, and said M is a number of constellation points;

The chooser unit is used for successively searching for successively, from last one deck, selects M to retain node as the signal candidate collection, calculates the metric of said signal candidate collection.

9. device according to claim 6 is characterized in that, said search unit comprises:

Obtain subelement, be used for obtaining the prior information of all layers corresponding bit from decoder at said tree structure by the bit layering;

The 3rd search subelement; Be used for prior information greater than the prior information weighted value of the node of second threshold value as metric; From M of upper strata M node retaining selectance value maximum 2M the node that node launches; M the node that said metric is maximum is as the retention node of current layer, and said M is a number of constellation points;

The chooser unit is used for successively searching for successively, from last one deck, selects M to retain node as the signal candidate collection, calculates the metric of said signal candidate collection.

10. according to any described device of claim 6-9, it is characterized in that the mapping mode that said quadrature amplitude modulation QAM sends the constellation point of signal comprises gray mappings or mapping naturally.

Images