CN101741519B - Sphere decoding method applied to multi-input multi-output channel - Google Patents

Abstract

The invention discloses a sphere decoding method applied to a multi-input multi-output (MIMO) channel, which comprises the following steps: receiving T signals transmitted through the MIMO channel; generating a first triangular matrix corresponding to the channel, and mapping the first triangular matrix to a real domain through a plurality of domains to acquire a second triangular matrix; solving a first zero-forcing soft-output solution, and acquire a plurality of better points P(1); acquiring a plurality of nth zero-forcing soft-output solutions corresponding to an nth estimation layer according to a plurality of better points P(n-1), and acquiring a plurality of better points P(n) according to part of Euclidean distance (PED) corresponding to a plurality of nth starlike points; and acquiring a plurality of 2Tth zero-forcing soft-output solution according to better points P(2T-1), accordingly acquiring a plurality of better points P(2T), and mapping the better points P(2T) corresponding to the minimum PED to the plurality of domains through the real domain to generate an optimal solution corresponding to the T signals.

Description

Be applied to the sphere decoding method of multi-input multi-output channel

Technical field

The present invention relates to a kind of sphere decoding method that is applied to multi-input multi-output channel, and be particularly related to a kind of sphere decoding method that is applied to multi-input multi-output channel that can reduce system complexity.

Background technology

Development along with wireless communication technology, the specification of radio area network is by single-input single-output (single-input single-output, SISO) 802.11a/b/g of pattern evolves to multiple-input and multiple-output (multi-input multi-output, MIMO) 802.11n of pattern, to meet transfer of data demand more at a high speed.Many transmission antennas of MIMO utilization and many reception antennas are to transmit respectively and to receive signal.Compared to traditional SISO system, mimo system provides the multiple parallel data flow under identical frequency band at one time simultaneously, and then the volume of transmitted data that provides multiple to grow up.

Mimo system must effectively utilize equivalent orthogonal characteristic passage on space to transmit the multiple parallel data flow in identical frequency band.Yet the equivalent orthogonal characteristic passage on space can be along with the geometric configuration and the element characteristic that transmit antenna and reception antenna, and the geometry of transmission path and statistical property and present decline in various degree.Therefore, one of important topic that input (signal detection) designs for mimo system.Generally speaking, be the complexity that promotes the signal estimating system, to reduce the error rate that receives signal.

In existing signal estimating and measuring method, maximum similarity (Maximum Likelihood, ML) rule can provide best receptivity, but can make the complexity of signal estimating system too high, and causes the difficulty on the hardware implementation.Therefore, develop sphere decoding (sphere decoding) method to approach the receptivity of maximum similarity rule, there is again suitable system complexity simultaneously.The framework of sphere decoding method comprises that rule is searched in depth-first (depth first) and breadth First (breadth first) is searched rule, these two kinds search rule all suitably setup parameter to reach the receptivity of maximum similarity rule, yet but must pay huge estimation complexity.Therefore, the estimation complexity that how effectively to reduce sphere decoding method just becomes at present one of important topic that in mimo system design institute must solution.

Summary of the invention

The present invention relates to a kind of sphere decoding method that is applied to multi-input multi-output channel, utilize the triangular matrix conversion rule of a plurality of territories to real number field, effectively reduce the estimation complexity of sphere decoding method, and can maintain the signal receiving performance of system.

According to a first aspect of the invention, a kind of sphere decoding method that is applied to multiple-input and multiple-output (MIMO) passage is proposed.The T that reception transmits via a MIMO passage signal, T is positive integer, the MIMO passage be take an access matrix as feature.Generation corresponds to one first triangular matrix of access matrix, and the first triangular matrix is mapped to real number field by a plurality of territories and obtains one second triangular matrix, and the second triangular matrix corresponds to one first estimation layer to a 2T estimation layer.Try to achieve and correspond to one first of the first estimation layer and force zero soft output solution, and obtain a plurality of better points P (1) of the first estimation layer.A plurality of better points P (n-1) according to (n-1) estimation layer, a plurality of n that obtain corresponding to a n estimation layer force zero soft output solution, each n forces zero soft output solution to correspond to separately the starlike point of a plurality of n, according to these n corresponding part of Euclidean distance of starlike point (PED), obtain a plurality of better points P (n) of n estimation layer, the positive integer that n is 2～(2T-1).These better points P (2T-1) according to (2T-1) estimation layer, a plurality of 2T that obtain corresponding to 2T estimation layer force zero soft output solution, and the corresponding a plurality of better points P (2T) that obtain 2T estimation layer, and map to a plurality of territories by these better points P (2T), corresponding to minimum PED person by real number field, correspond to an optimum solution of T signal with generation.

For foregoing of the present invention can be become apparent, a preferred embodiment cited below particularly, and coordinate accompanying drawing, be described in detail below:

The accompanying drawing explanation

Fig. 1 illustrates the flow chart according to the sphere decoding method that is applied to the MIMO passage of the preferred embodiment of the present invention.

[main element symbol description]

S100～S140: steps flow chart

Embodiment

The present invention proposes a kind of sphere decoding method that is applied to multi-input multi-output channel (MIMO), utilize the triangular matrix conversion rule of a plurality of territories (complex domain) to real number field (real domain), carry out the Q-R decomposition in a plurality of territories, and try to achieve a plurality of better points corresponding to each estimation layer in real number field.Thus, the estimation complexity of sphere decoding method can be effectively reduced, and the signal receiving performance of system can be maintained.

The sphere decoding method that is applied to the MIMO passage of the present embodiment, comprise the following steps: that a. receives T the signal transmitted via the MIMO passage, and T is positive integer, and the MIMO passage be take an access matrix as feature; B. produce and correspond to one first triangular matrix of access matrix, and the first triangular matrix is mapped to real number field by a plurality of territories and obtain one second triangular matrix, the second triangular matrix corresponds to one first estimation layer to a 2T estimation layer; C. try to achieve and correspond to one first of the first estimation layer and force zero soft output (zero-forcing soft-output) to separate, and obtain a plurality of better points P (1) of the first estimation layer; D. estimate a plurality of better points P (n-1) of layer according to one (n-1), a plurality of n that obtain corresponding to a n estimation layer force zero soft output solution, each n forces zero soft output solution to correspond to separately the starlike point of a plurality of n, according to the corresponding part of Euclidean distance of the starlike point of these n (partial Euclidean distance, PED), obtain a plurality of better points P (n) of n estimation layer, the positive integer that n is 2～(2T-1); And e. is according to these better points P (2T-1) of (2T-1) estimation layer, a plurality of 2T that obtain corresponding to 2T estimation layer force zero soft output solution, and the corresponding a plurality of better points P (2T) that obtain 2T estimation layer, and map to a plurality of territories by these better points P (2T), corresponding to minimum PED person by real number field, correspond to an optimum solution of T signal with generation.

Further be described in detail as follows now.Please refer to Fig. 1, it illustrates the flow chart according to the sphere decoding method that is applied to the MIMO passage of the preferred embodiment of the present invention.At first, in step S100, receive T the signal transmitted via a MIMO passage, this MIMO passage be take an access matrix as feature.Preferably, be to utilize R ' root reception antenna to receive the T-root via the MIMO passage to transmit T the signal that antenna is exported, R ' is positive integer.This access matrix is a R ' * T access matrix H.Mimo system with time domain or frequency domain representation can be defined as follows:

r＝Hx+n (1)

Wherein, the matrix r of R ' * 1 represents the signal that R ' root reception antenna receives, and the matrix x of T * 1 represents that the T root transmits the signal that antenna is exported, and the matrix n of R ' * 1 represents noise.

Then, in step S110, access matrix H is carried out to a Q-R and decompose to produce one first triangular matrix, and this first triangular matrix is mapped to real number field by a plurality of territories and obtain one second triangular matrix, this second triangular matrix corresponds to one first estimation layer to a 2T estimation layer.Decompose and can obtain by above-mentioned Q-R:

H＝QR

Wherein, the matrix that Q is a R ' * T, and the first triangular matrix that R is one T * T.This first triangular matrix R is for example a upper triangular matrix, but is not limited to this.After step S110, another equation that can obtain this mimo system according to equation (1) is as follows:

y＝Q ^Hr＝Rx+Q ^Hn (2)

Wherein, Q ^hwith the product of Q be unit matrix I.If the noise item Q by equation (2) ^hn ignores, and the first triangular matrix R is while for example being upper triangular matrix, and equation (2) can be unfolded as follows by matrix form:

(3)

Wherein, y _p,

and

be known, wherein

for r _pqreal part,

for r _pqimaginary part, the positive integer that p is 1～T, the positive integer that q is 1～T.Then, the first triangular matrix R is mapped to real number field and obtained one second triangular matrix R by a plurality of territories, wherein, corresponding to the first triangular matrix R is upper triangular matrix, and the second triangular matrix R is also upper triangular matrix.Thus, equation (3) can be changed as follows:

(4)

Wherein, equation (3) and equation (4) are launched and compare, can obtain the coefficient in equation (4) $c_{\mathrm{pq}}^{\left(1\right)}~c_{\mathrm{pq}}^{\left(4\right)}$ As follows:

$\left\{\begin{array}{c}{c}_{\mathrm{pq}}^{\left(1\right)}=1\\ c_{\mathrm{pq}}^{\left(2\right)}=-1\\ c_{\mathrm{pq}}^{\left(3\right)}=1\\ c_{\mathrm{pq}}^{\left(4\right)}=1\end{array}\right.$ And $\left\{\begin{array}{c}{r}_{\mathrm{pq}}^{\left(1\right)}=r_{\mathrm{pq}}^{\left(r\right)}\\ r_{\mathrm{pq}}^{\left(2\right)}=r_{\mathrm{pq}}^{\left(i\right)}\end{array}\right.,$ $\mathrm{if}{x}_q^{\left(1\right)}=x_q^{\left(r\right)}$ And $y_p^{\left(1\right)}=y_p^{\left(r\right)}$ , or

$\left\{\begin{array}{c}{c}_{\mathrm{pq}}^{\left(1\right)}=-1\\ c_{\mathrm{pq}}^{\left(2\right)}=1\\ c_{\mathrm{pq}}^{\left(3\right)}=1\\ c_{\mathrm{pq}}^{\left(4\right)}=1\end{array}\right.$ And $\left\{\begin{array}{c}{r}_{\mathrm{pq}}^{\left(1\right)}=r_{\mathrm{pq}}^{\left(i\right)}\\ r_{\mathrm{pq}}^{\left(2\right)}=r_{\mathrm{pq}}^{\left(r\right)}\end{array}\right.,$ $\mathrm{if}{x}_q^{\left(1\right)}=x_q^{\left(i\right)}$ And $y_p^{\left(1\right)}=y_p^{\left(r\right)}$ , or

$\left\{\begin{array}{c}{c}_{\mathrm{pq}}^{\left(1\right)}=1\\ c_{\mathrm{pq}}^{\left(2\right)}=1\\ c_{\mathrm{pq}}^{\left(3\right)}=1\\ c_{\mathrm{pq}}^{\left(4\right)}=-1\end{array}\right.$ And $\left\{\begin{array}{c}{r}_{\mathrm{pq}}^{\left(1\right)}=r_{\mathrm{pq}}^{\left(i\right)}\\ r_{\mathrm{pq}}^{\left(2\right)}=r_{\mathrm{pq}}^{\left(r\right)}\end{array}\right.,$ $\mathrm{if}{x}_q^{\left(1\right)}=x_q^{\left(r\right)}$ And $y_p^{\left(1\right)}=y_p^{\left(i\right)}$ , or

$\left\{\begin{array}{c}{c}_{\mathrm{pq}}^{\left(1\right)}=1\\ c_{\mathrm{pq}}^{\left(2\right)}=1\\ c_{\mathrm{pq}}^{\left(3\right)}=-1\\ c_{\mathrm{pq}}^{\left(4\right)}=1\end{array}\right.$ And $\left\{\begin{array}{c}{r}_{\mathrm{pq}}^{\left(1\right)}=r_{\mathrm{pq}}^{\left(r\right)}\\ r_{\mathrm{pq}}^{\left(2\right)}=r_{\mathrm{pq}}^{\left(i\right)}\end{array}\right.,$ $\mathrm{if}{x}_q^{\left(1\right)}=x_q^{\left(i\right)}$ And $y_p^{\left(1\right)}=y_p^{\left(i\right)}$

Wherein, the present embodiment will be by known

and

value, obtain

optimum solution.

In equation (4), correspond to equation $y_T^{\left(2\right)}=r_{\mathrm{TT}}{x}_T^{\left(2\right)}$ Be defined as the first estimation layer, correspond to equation $y_T^{\left(1\right)}=r_{\mathrm{TT}}{x}_T^{\left(1\right)}$ Be defined as the second estimation layer, correspond to

equation sequentially be defined as the 3rd estimation layer～(2T-2) and estimate layer, correspond to

equation $y_1^{\left(2\right)}=c_{1T}^{\left(4\right)}\·r_{1T}^{\left(1\right)}{x}_T+...+r_{11}{x}_1^{\left(2\right)}$ Be defined as (2T-1) estimation layer, correspond to

equation $y_1^{\left(1\right)}=c_{1T}^{\left(2\right)}\·r_{1T}^{\left(2\right)}{x}_T+...+r_{11}{x}_1^{\left(1\right)}$ Be defined as 2T estimation layer.Wherein, each estimation layer comprises M starlike point (constellation point), and M is positive integer.

Then, in step S120, try to achieve and correspond to the first estimation layer

one first force zero soft output solution, and first force zero soft output solution to obtain a plurality of better points P (1) of the first estimation layer according to this.Wherein, corresponding to the first estimation layer, be to separate $y_T^{\left(2\right)}=r_{\mathrm{TT}}{x}_T^{\left(2\right)}$ And try to achieve

first force zero soft output solution

first force real part and the imaginary part of zero soft defeated solution all may be with floating number.Carry out computing in real number field due to the present embodiment, thus according to

the first distance of forcing the one dimension between zero soft defeated solution, can find the most approaching

first force a plurality of first of zero soft defeated solution starlikely to using as a plurality of better points P (1).Therefore must not calculate in fact a plurality of better points P (1) that the PED that reaches M time can find the first estimation layer in the first estimation layer, make the reduced complexity of system.

Come again, in step S130, K better points P (n-1) according to (n-1) estimation layer, a plurality of n that obtain corresponding to n estimation layer force zero soft output solution, each n forces zero soft output solution to correspond to separately the starlike point of a plurality of n, and, according to the corresponding PED of the starlike point of these n, obtains a plurality of better points P (n) of n estimation layer, the positive integer that n is 2～(2T-1), the K value is less than M.

In step S130, when the value of n is 2, a plurality of better points P (1) according to the first estimation layer, separate $y_T^{\left(1\right)}=r_{\mathrm{TT}}{x}_T^{\left(1\right)}$ And obtain corresponding to a plurality of of the second estimation layer

second force zero soft output solution.Owing to carrying out computing in real number field, therefore according to a plurality of starlike points and a plurality of

the second distance of forcing the one dimension of zero soft output solution, can find rapidly each

second force zero soft each self-corresponding a plurality of second starlike point of output solution, this each the self-corresponding a plurality of second starlike point is preferably each near the second a plurality of second starlike point of forcing zero soft output solution.Each second number of forcing zero soft output to separate the corresponding second starlike point is identical or different.

Then, obtain the corresponding PED value of each the second starlike point.Obtain a plurality of better points P (2) of the second estimation layer in these second starlike points, and these a plurality of better points P (2) are passed to the 3rd estimation layer.Wherein, any the corresponding PED in these a plurality of better points P (2) is less than any the corresponding PED in other the second starlike point.

When the n value is less than or equal to (2T-2), the n value is added to 1.Thus, according to a plurality of better points P (n-1) of (n-1) estimation layer, a plurality of n that obtain corresponding to n estimation layer force zero soft output solution.Owing to carrying out computing in real number field, therefore can find rapidly each n to force the starlike point of zero soft output solution each self-corresponding a plurality of n of institute according to the distance of one dimension.The number that each n forces zero soft output to separate the starlike point of corresponding n is identical or different.From the starlike point of these n, obtain a plurality of better points P (n) of n estimation layer.Wherein, any the corresponding PED in these better points P (n) is less than any the corresponding PED in other the starlike point of n.Thus, must not calculate nearly K the better points P (n) that KM PED can find n estimation layer in n estimation layer, make the reduced complexity of system.

Afterwards, in step S140, a plurality of better points P (2T-1) according to (2T-1) estimation layer, separate $y_1^{\left(1\right)}=c_{1T}^{\left(2\right)}\·r_{1T}^{\left(2\right)}{x}_T+...+r_{11}{x}_1^{\left(1\right)}$ Estimate a plurality of of layer and obtain corresponding to 2T

2T force zero soft output solution.Owing to carrying out computing in real number field, thus according to and each

2T force the distance of the one dimension between zero soft defeated solution, can find and approach each most

2T force the starlike point of a plurality of 2T between zero soft defeated solution, the starlike point of these 2T is better points P (2T).Afterwards, will in these better points P (2T), correspond to minimum PED person, foundation

and

with

and

between man-to-man corresponding relation, map to a plurality of territories by real number field, correspond to an optimum solution of T signal with generation, that is be the optimum solution of matrix x.Thus, must not calculate nearly a plurality of better points P (2T) that M PED can find 2T estimation layer in 2T estimation layer, make the reduced complexity of system.

Lift now 2 signals that utilize 2 reception antennas to export via 2 transmission antennas of MIMO passage reception and explain for example, so be not limited to this.When receiving 2 signals that 2 transmission antennas export via the MIMO passage, mimo system can be unfolded as follows by matrix form:

$\left[\begin{array}{c}{y}_1\\ y_2\end{array}\right]=\left[\begin{array}{cc}{r}_{11}& r_{12}^{\left(r\right)}+i\·r_{12}^{\left(i\right)}\\ 0& r_{22}\end{array}\right]\left[\begin{array}{c}{x}_1\\ x_2\end{array}\right]$ (5)

If $\left\{\begin{array}{c}{c}_{\mathrm{pq}}^{\left(1\right)}=1\\ c_{\mathrm{pq}}^{\left(2\right)}=-1\\ c_{\mathrm{pq}}^{\left(3\right)}=1\\ c_{\mathrm{pq}}^{\left(4\right)}=1\end{array}\right.$ And $\left\{\begin{array}{c}{r}_{\mathrm{pq}}^{\left(1\right)}=r_{\mathrm{pq}}^{\left(r\right)}\\ r_{\mathrm{pq}}^{\left(2\right)}=r_{\mathrm{pq}}^{\left(i\right)}\end{array}\right.,$ $\mathrm{if}{x}_q^{\left(1\right)}=x_q^{\left(r\right)}$ And $y_p^{\left(1\right)}=y_p^{\left(r\right)}$ ，

Equation (5) can be changed as follows:

$\left[\begin{array}{c}{y}_1^{\left(r\right)}\\ y_1^{\left(i\right)}\\ y_2^{\left(r\right)}\\ y_2^{\left(i\right)}\end{array}\right]=\left[\begin{array}{cccc}{r}_{11}& 0& r_{12}^{\left(r\right)}& -r_{12}^{\left(i\right)}\\ 0& r_{11}& r_{12}^{\left(i\right)}& r_{12}^{\left(r\right)}\\ 0& 0& r_{22}& 0\\ 0& 0& 0& r_{22}\end{array}\right]\left[\begin{array}{c}{x}_1^{\left(r\right)}\\ x_1^{\left(i\right)}\\ x_2^{\left(r\right)}\\ x_2^{\left(i\right)}\end{array}\right]$ (6)

In equation (6), correspond to

equation $y_2^{\left(i\right)}=r_{22}{x}_2^{\left(i\right)}$ Be defined as the first estimation layer, correspond to

equation $y_2^{\left(r\right)}=r_{22}{x}_2^{\left(r\right)}$ Be defined as the second estimation layer, correspond to

equation $y_1^{\left(i\right)}=r_{12}^{\left(r\right)}{x}_2^{\left(i\right)}+r_{12}^{\left(i\right)}{x}_2^{\left(r\right)}+r_{11}{x}_1^{\left(i\right)}$ Be defined as the 3rd estimation layer, correspond to

equation $y_1^{\left(r\right)}=-r_{12}^{\left(i\right)}{x}_2^{\left(i\right)}+r_{12}^{\left(r\right)}{x}_2^{\left(r\right)}+r_{11}{x}_1^{\left(r\right)}$ Be defined as the 4th estimation layer, sequentially by the first estimation layer to the four estimation layers, separate corresponding equation, find and correspond to minimum PED person, can find

be for example (a ₁, b ₁, a ₂, b ₂).Then, by (a ₁, b ₁, a ₂, b ₂) map to a plurality of territories by real number field and can obtain (x ₁, x2 ₎for (a ₁+ ib ₁, a ₂+ ib ₂), this correspond to the optimum solution of 2 signals.

The disclosed sphere decoding method that is applied to the MIMO passage of the above embodiment of the present invention, be to propose the triangular matrix conversion rule of simple a plurality of territories to real number field, and be able to adopt the Q-R in a plurality of territories that complexity is lower to decompose at the leading portion of the flow process that solves T signal.In addition, because the first triangular matrix R maps to real number field by a plurality of territories, be the second triangular matrix R, therefore can in real number field, to each estimation layer, be estimated a plurality of better points in the hope of correspondence at the back segment of the flow process that solves T signal.Thus, when each estimation layer is estimated, need only carry out simple ordering action, and need only calculate the corresponding PED of the starlike point of fewer purpose, therefore can effectively reduce the estimation complexity of sphere decoding method.

In addition, if while in real number field, each estimation layer being estimated, in conjunction with SE (Schnorr&Euchner) enumeration method, further originally must carry out complicated ordering action in a plurality of territories and can be reduced to ordering action simple in real number field, can significantly simplify the complexity while being enumerated.Therefore the receiving terminal of the sphere decoding method of the MIMO passage of use the present embodiment can have dynamical signal transmission and signal decoding, can reach the receptivity of approximate maximum similarity rule, and reduce the advantage of hardware complexity and the minimizing chip area of system.

In sum, although the present invention with a preferred embodiment openly as above, so it is not in order to limit the present invention.Those skilled in the art under the present invention, without departing from the spirit and scope of the present invention, when being used for a variety of modifications and variations.Therefore, protection scope of the present invention is as the criterion when looking the appended claims person of defining.

Claims (3)

1. a sphere decoding method that is applied to multiple-input and multiple-output MIMO passage comprises:

A. receive T the signal transmitted via this MIMO passage, T is positive integer, and this MIMO passage be take an access matrix as feature;

B. produce and correspond to one first triangular matrix of this access matrix, and this first triangular matrix is mapped to real number field by a plurality of territories and obtain one second triangular matrix, this second triangular matrix corresponds to one first estimation layer to a 2T estimation layer;

C. try to achieve and correspond to one first of this first estimation layer and force zero soft output solution, and obtain a plurality of better points P (1) of this first estimation layer;

D. estimate a plurality of better points P (n-1) of layer according to one (n-1), a plurality of n that obtain corresponding to a n estimation layer force zero soft output solution, each n forces zero soft output solution to correspond to separately the starlike point of a plurality of n, according to the corresponding part of Euclidean distance PED of the starlike point of these n, obtain a plurality of better points P (n) of this n estimation layer, the positive integer that n is 2～(2T-1); And

E. estimate these better points P (2T-1) of layer according to this (2T-1), a plurality of 2T that obtain corresponding to this 2T estimation layer force zero soft output solution, and the corresponding a plurality of better points P (2T) that obtain this 2T estimation layer, and map to a plurality of territories by these better points P (2T), corresponding to minimum PED person by real number field, correspond to an optimum solution of this T signal with generation

Wherein, in this step b, this access matrix is carried out to a Q-R and decompose to produce this first triangular matrix.

2. the method for claim 1, wherein this first triangular matrix and this second triangular matrix are upper triangular matrix.

3. method as claimed in claim 2, wherein this first estimation layer corresponds to the 2T row element of this second triangular matrix.