CN101741522B - 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: arranging a plurality of starlike points of an nth estimation layer by an enumeration method and according to the nth estimation layer corresponding to a matrix of the MIMO channel, and defining at least one nth subset of the nth estimation layer; taking K starlike points serving as better points out from the nth subset, and selecting Kn better points of the nth estimation layer through K better points of all nth subsets; transferring K1 better points to a second estimation layer through a first estimation layer, transferring K2 better points to a third estimation layer through the second estimation layer,..., transferring K(T-1) better points to a T estimation layer through a (T-1) estimation layer; and determining an optimal solution according to KT better points of the nth estimation layer, wherein at least one of numbers of K and K1 to KT is determined by the matrix characteristic of the MIMO channel.

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 is to satisfy 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 the space to transmit the multiple parallel data flow in identical frequency band.Yet the equivalent orthogonal characteristic passage on the 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 in various degree decline.Therefore, one of important topic of designing for mimo system of input (signal detection).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 with the receptivity near the maximum similarity rule, have again suitable system complexity simultaneously.In sphere decoding method, the K-Best sphere decoding method of breadth First (breadth first) is because having fixedly treating capacity (throughput) and fixedly complexity, therefore be very suitable for the hardware implementation.

For the K-Best sphere decoding method, the height of its receptivity is decided by the selection of K value.If the K value is larger, signal receiving performance is higher, but can cause system complexity to increase.If the K value is less, system complexity reduces, but signal receiving performance can step-down.Therefore, the K-Best sphere decoding method must make trade-offs to determine the K value between system complexity and receptivity.In addition, in the K-Best sphere decoding method, each estimation layer (detection layer) comprises M starlike point, and M is positive integer.The K-Best sphere decoding method accepts at each estimation layer K better points estimating layer from last, and must a first calculating K M part of Euclidean distance (partial Euclidean distance), then be delivered to next via K the better points that present estimation layer is tried to achieve in sequence (sorting) and estimate layer.

If the signal number that transmits via the MIMO passage is T, T is positive integer, the K-Best sphere decoding method corresponds to T estimation layer under a plurality of territories (complex domain), and the K-Best sphere decoding method must calculate KM+K Euclidean distance of M+ (T-2) altogether, and carries out the action of a minor sort at each estimation layer.Yet, when serious or orthogonal characteristic is not enough when the decline situation of MIMO passage, for the mimo system of high-order, must improve the K value with the signal receiving performance of elevator system.Thus, the complexity that but can cause system hurriedly increases with the ratio of (T-2) M, makes the area of chip processor become large, power consumption increases and data processing amount descends.

Summary of the invention

The present invention relates to a kind of sphere decoding method that is applied to multi-input multi-output channel, based on the SE enumeration method, according to the corresponding signal to noise ratio of each estimation layer (Signal to Noise Ratio, SNR), obtain corresponding better points from the starlike point of part of each estimation layer, reduce the estimation complexity of sphere decoding method, and can keep 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.Correspond to a n estimation layer according to the MIMO access matrix, arrange a plurality of starlike point of n estimation layer with an enumeration method, and define at least one n subclass of n estimation layer, wherein the MIMO access matrix corresponds to the 1st～the T estimation layer, n estimation layer and K _nCorrespond to respectively the 2nd～the (T-1) estimation layer and K ₂～K _(T-1)At least one of them, K _nBe positive integer.Take out K individual starlike as better points from the n subclass, K is positive integer, K _nSummation for all n subclass better points number K.Transmit K by the 1st estimation layer ₁Individual better points to the 2 estimation layers, the 2nd estimation layer transmits K ₂Individual better points to the 3 estimation layers ..., (T-1) estimation layer transmits K _(T-1)Individual better points to the T estimation layer.K according to T estimation layer _TIndividual better points determines an optimum solution.Wherein, K and K ₁～K _TThe number size one of them is determined by MIMO access matrix characteristic at least.

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

Description of drawings

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

Fig. 2 illustrates the flow chart according to the sphere decoding method that is applied to the MIMO passage of first embodiment of the invention.

Fig. 3 illustrates the star-plot of an estimation layer of sphere decoding method.

Fig. 4 illustrates according to the real part candidate value that corresponds to the first estimation layer of the first embodiment of the invention schematic diagram to an example of imaginary part candidate value.

Fig. 5 illustrates according to the real part candidate value that corresponds to the first estimation layer of first embodiment of the invention another routine schematic diagram to the imaginary part candidate value.

Fig. 6 illustrates according to the real part candidate value that corresponds to the second estimation layer of the first embodiment of the invention schematic diagram to an example of imaginary part candidate value.

The T that corresponds to that Fig. 7 illustrates according to first embodiment of the invention estimates the real part candidate value of layer to the schematic diagram of an example of imaginary part candidate value.

Fig. 8 illustrates the flow chart according to the sphere decoding method that is applied to the MIMO passage of second embodiment of the invention.

Fig. 9 illustrates the schematic diagram to an example of the second starlike point according to the better points P (1) that corresponds to the second estimation layer of second embodiment of the invention.

Figure 10 illustrates another routine schematic diagram to the second starlike point according to the better points P (1) that corresponds to the second estimation layer of second embodiment of the invention.

Figure 11 illustrates the schematic diagram to an example of the 3rd starlike point according to the better points P (2) after the arrangement that corresponds to the 3rd estimation layer of second embodiment of the invention.

Figure 12 illustrates the schematic diagram to an example of the starlike point of 2T according to the better points P (2T-1) after the arrangement that corresponds to 2T estimation layer of second embodiment of the invention.

[main element symbol description]

402,404,406,1102,1104,1106,1108: the zone

601～610: the shadow region

S100～S130, S200～S240, S800～S850: process step

Embodiment

The present invention proposes a kind of sphere decoding method that is applied to multiple-input and multiple-output (MIMO) passage, respectively in a plurality of territories and real number field (real domain), based on SE (Schnorr ﹠amp; Euchner) enumeration method, according to the corresponding signal to noise ratio of each estimation layer (Signal to Noise Ratio, SNR), can obtain a plurality of better points of corresponding this estimation layer from the starlike point of part (constellation point) of each estimation layer, reduce the estimation complexity of sphere decoding method, and can keep the signal receiving performance of mimo system.

Please refer to Fig. 1, it illustrates the flow chart according to the sphere decoding method that is applied to the MIMO passage of preferred embodiment of the present invention.In step S100, correspond to a n estimation layer according to the MIMO access matrix, arrange a plurality of starlike point of n estimation layer with an enumeration method, and define at least one n subclass of n estimation layer.Wherein, this enumeration method is for example the SE enumeration method.Wherein, the MIMO access matrix corresponds to the 1st～the T estimation layer, n estimation layer and K _nCorrespond to respectively the 2nd～the (T-1) estimation layer and K ₂～K _(T-1)At least one of them, K _nBe positive integer.In step S110, take out K individual starlike as better points from the n subclass, and selected the K of n estimation layer by K better points of all n subclass _nIndividual better points, K is positive integer, K _nBe the positive integer less than or equal to K better points number summation of all n subclass.In step S120, transmit K by the 1st estimation layer ₁Individual better points to the 2 estimation layers, the 2nd estimation layer transmits K ₂Individual better points to the 3 estimation layers ..., (T-1) estimation layer transmits K _(T-1)Individual better points to the T estimation layer.In step S130, according to the K of T estimation layer _TIndividual better points determines an optimum solution.Wherein, K and K ₁～K _TThe number size one of them is determined by MIMO access matrix characteristic at least.Wherein MIMO access matrix characteristic refers to that the MIMO access matrix corresponds to the reliability of forcing zero soft output (zero-forcingsoft-output) to separate of each estimation layer.Wherein the reliability of each estimation layer depends on signal to noise ratio (SNR) intensity of each estimation layer.

Next lift now that above-mentioned sphere decoding method is applied to respectively a plurality of territories and real number field is that example explains, and explain as the SE enumeration method as example take enumeration method, so be not limited to this.Wherein, the triangular matrix that corresponds to access matrix explains as an example of upper triangular matrix example, and makes the reliability of T estimation layer sequentially successively decrease, that is K ₁＞=K ₂＞=...＞=K _T, so be not limited to this, K- _nDetermined by MIMO access matrix characteristic in fact, that is K _n--The signal to noise ratio intensity that depends in fact n estimation layer.

The first embodiment

The sphere decoding method that is applied to the MIMO passage of the present embodiment is used in a plurality of territories, and it comprises the following steps: that a. receives T the signal that transmits via the MIMO passage, and T is positive integer, and the MIMO passage is take an access matrix as feature; B. produce a triangular matrix that corresponds to access matrix, triangular matrix corresponds to one first estimation layer to a T estimation layer, wherein the corresponding signal to noise ratio of j estimation layer is that j is 1 positive integer to (T-1) more than or equal to the corresponding signal to noise ratio of (j+1) 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 the K of the first estimation layer based on the SE enumeration method ₁Individual better points P (1); D. estimate the K of layer according to (n-1) _n-1Individual better points P (n-1) obtains corresponding to the K that n estimates layer _n-1Individual n forces zero soft output solution, and based on the SE enumeration method, selects K _n-1Individual n forces the 1 starlike point of the corresponding a plurality of n of soft output solution, and according to the corresponding part of Euclidean distance of the starlike point of these n (partial Euclideandistance, PED), obtains the K of n estimation layer _nIndividual better points P (n), n are 2～(T-1) positive integer; And e. is according to the K of (T-1) estimation layer _T-1Individual better points P (T-1) obtains corresponding to the K that T estimates layer _T-1Individual T forces zero soft output solution, and based on the SE enumeration method, obtains the K of T estimation layer _TIndividual better points P (T), and according to K _TCorrespond to minimum PED person in individual better points P (T), produce an optimum solution that corresponds to T signal.Wherein, K ₁, K ₂... and K _TBe positive integer, and K ₁＞=K ₂＞=...＞=K _T

Further be described in detail as follows now.Please refer to Fig. 2, it illustrates the flow chart according to the sphere decoding method that is applied to the MIMO passage of first embodiment of the invention.At first, in step S200, receive T the signal that transmits via a MIMO passage, this MIMO passage is 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 S210, access matrix H is carried out a Q-R decompose to produce a triangular matrix, this triangular matrix corresponds to one first estimation layer to a T estimation layer.Preferably, the first estimation layer～T estimation layer corresponds to respectively in triangular matrix, matrix element be zero element number maximum, inferior many ..., a minimum column element.Decompose and can get by above-mentioned Q-R:

H＝QR

Wherein, Q is the matrix of a R ' * T, and R is the triangular matrix of one T * T.This triangular matrix R is for example a upper triangular matrix, but is not limited to this.After step S210, 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 with equation (2) ^HN ignores, and triangular matrix R is when for example being upper triangular matrix, and equation (2) can be unfolded as follows by matrix form:

(3)

Wherein, y _pAnd r _pqBe known, p is the positive integer of 1～T, and q is the positive integer of 1～T.The present embodiment will be by known y _pAnd r _pqValue, obtain x ₁～x _TOptimum solution.

In equation (3), correspond to x _TEquation y _T=r _TTx _TBe defined as the first estimation layer, correspond to x _T-1..., x ₂Equation sequentially be defined as the second estimation layer～(T-1) and estimate layer, correspond to x ₁Equation y ₁=r _1Tx _T+ ...+r ₁₁x ₁Be defined as T estimation layer.Wherein, each estimation layer comprises M starlike point, and M is positive integer.

Please refer to Fig. 3, it illustrates the star-plot of an estimation layer of sphere decoding method.In Fig. 3, comprise that with each estimation layer 64 starlike points (representing take hollow dots) explain as example, but be not limited to this.The K-Best sphere decoding method produces K corresponding to n estimation layer _nIndividual better points, n is positive integer and the K of 1～T _nLess than M.

Then, in step S220, try to achieve the x that corresponds to the first estimation layer _TOne first force zero soft output solution, and obtain the K of the first estimation layer based on the SE enumeration method ₁Individual better points P (1).Wherein, corresponding to the first estimation layer, be to separate y _T=r _TTx _TAnd try to achieve x _TFirst force zero soft output solution y _T/ r _TT, x _TFirst to force real part and the imaginary part of zero soft defeated solution may with floating number, be all for example the some Z (1.4,2.3) in Fig. 3.And in following step, will find out near the K of Z point ₁Individual starlike with as K ₁Individual better points.

Above-mentioned SE enumeration method foundation and the distance (distance of the distance of real part axle or imaginary part axle) of the one dimension between a desired value sequentially list near the integer value of this desired value from small to large.The numerical value 1.4 of lifting the real part that Z order is example as desired value, can sequentially obtain apart from the value 1,3 ,-1,5 ,-3,7 ,-5 of this desired value by real part axle closely extremely far away based on the SE enumeration method ... etc. integer value.Because the algorithm of SE enumeration method is very simple, therefore can effectively reduce system's operand.Wherein, in the coordinate figure of Fig. 3, real part axle and imaginary part axle represent respectively in the QAM modulation of mimo system, the amplitude size of the carrier wave of two quadratures.

Step S220 arranges in fact a plurality of starlike point of the first estimation layer with the SE enumeration method, and defines one first subclass of the first estimation layer.Wherein, the included starlike some candidate's scope of the first subclass is ordered by real part axle candidate scope M1 and imaginary part axle candidate scope M2.In step S220, for x _TFirst force zero soft output solution, obtain M based on the SE enumeration method ₁Individual real part candidate value and M ₂Individual imaginary part candidate value, M ₁And M ₂Be positive integer, M ₁And M ₂All less than or equal to

, M ₁With M ₂Product more than or equal to K ₁Then, make M ₁Individual real part candidate value and M ₂The value of corresponding a plurality of first starlike point of individual imaginary part candidate value is x _TDifference substitution r _TTx _T, and resulting result is defined as respectively y ' _TSo, the corresponding PED of each first starlike point be corresponding (y ' _T-y _T) ²Square root.Can obtain the K of the first estimation layer in these first starlike points ₁Individual better points P (1), wherein, this K ₁Any corresponding PED in individual better points P (1) is less than any the corresponding PED in other the first starlike point.Thus, must not calculate nearly the K that M PED can find the first estimation layer in the first estimation layer ₁Individual better points P (1) makes the reduced complexity of system.

Then, in step S230, according to the K of (n-1) estimation layer _n-1Individual better points P (n-1) obtains corresponding to the K that n estimates layer _(n-1)Individual x _T-n+1N force zero soft output solution, and based on the SE enumeration method, select K _(n-1)Individual n forces the 1 starlike point of the corresponding a plurality of n of soft output solution, and according to the corresponding PED of the starlike point of these n, obtains the K of n estimation layer _nIndividual better points P (n), n are 2～(T-1) positive integer.

Step S230 arranges in fact a plurality of starlike point of n estimation layer with the SE enumeration method, and defines a plurality of n subclass of n estimation layer.In step S230, when the value of n is 2, according to the K of the first estimation layer ₁Individual better points P (1) separates y _T-1=r _{(T-1) T}x _T+ r _{(T-1) (T-1)}x _T-1And obtain corresponding to the K of the second estimation layer ₁Individual x _T-1Second force zero soft output solution, and based on the SE enumeration method, for K ₁Individual x _T-1Second force zero soft output solution to obtain separately M ₁Individual real part candidate value and M ₂Individual imaginary part candidate value.Each second forces zero soft output solution corresponding M separately ₁Value and a M ₂Value, each second forces zero soft output to separate each self-corresponding M ₁Value and M ₂Value be identical or different.In these real part candidate values and the corresponding starlike point of imaginary part candidate value, corresponding to K ₁Individual x _T-1Second force zero soft output solution can obtain respectively K ₁Individual the second subclass, this K ₁Individual the second subclass takes out K the second starlike point separately, and wherein each second corresponding K value of subclass is identical or different.Lead because of in the advantage that adopts the SE enumeration method, this K ₁Individual the second corresponding K value of subclass can be descending.

Then, making respectively the value of these the second starlike points is x _T-1, with its value substitution r with corresponding better points P (1) _{(T-1) T}x _T+ r _{(T-1) (T-1)}x _T-1, and resulting result is defined as y ' _T-1So, the corresponding PED of each second starlike point be corresponding (y ' _T-y _T) ²+ (y ' _T-1-y _T-1) ²Square root.Obtain the K of the second estimation layer in these second starlike points ₂Individual better points P (2), and with this K ₂Individual better points P (2) is passed to the 3rd estimation layer.Wherein, this K ₂Any corresponding PED in individual better points P (2) is less than any the corresponding PED in other the second starlike point.

During less than or equal to (T-2), the n value is added 1 when the n value.Thus, estimate the K of layer according to (n-1) _(n-1)Individual better points P (n-1) separates y _T-n+1=r _{(T-n+1) T}x _T+ ...+r _{(T-n+1) (T-n+1)}x _T-n+1And obtain corresponding to the K of n estimation layer _(n-1)Individual x _T-n+1N force zero soft output solution, and based on the SE enumeration method, each x _T-n+1N force zero a soft output solution corresponding n subclass separately, that is correspond to separately M ₁Individual real part candidate value and M ₂Individual imaginary part candidate value.Each n forces zero soft output solution corresponding M separately ₁Value and a M ₂Value.Wherein, for each x _T-n+1N force zero soft output solution, K, M ₁And M ₂Can be different.In fact, K, M ₁And M ₂At least one of them is determined by MIMO access matrix characteristic.Preferably, the K of n estimation layer _(n-1)Individual x _T-n+1N force zero soft output solution to distinguish the K of correspondence _(n-1)Individual n subclass corresponds to respectively different K values according to the PED of correspondence.

From the starlike point of a plurality of n that corresponds to these real part candidate values and these imaginary part candidate values, obtain the K of n estimation layer _nIndividual better points P (n).Wherein, the corresponding PED of the starlike point of each n be corresponding (y ' _T-y _T) ²+ (y ' _T-1-y _T-1) ²+ ...+(y ' _T-n+1-y _T-n+1) ²Square root.Wherein, this K _nAny corresponding PED in individual better points P (n) is less than any the corresponding PED in the starlike point of other n.Thus, must not calculate nearly K in n estimation layer _(n-1)M PED can find the K of n estimation layer _nIndividual better points P (n) makes the reduced complexity of system.

Afterwards, in step S240, according to the K of (T-1) estimation layer _(T-1)Individual better points P (T-1) separates y ₁=r _1Tx _T+ ...+r ₁₁x ₁And obtain corresponding to the K of T estimation layer _(T-1)Individual x ₁T force zero soft output solution, and based on the SE enumeration method, obtain the K of T estimation layer _TIndividual better points P (T), and according to K _TCorrespond to minimum PED person in individual better points P (T), produce and correspond to one of T signal optimum solution, that is be the optimum solution of matrix x.Wherein, this K _TIn individual better points P (T), each self-corresponding PED be corresponding (y ' _T-y _T) ²+ (y ' _T-1-y _T-1) ²+ ...+(y ' ₁-y ₁) ²Square root.Thus, must not calculating nearly in T estimation layer, M PED can find T to estimate the K of layer _TIndividual better points P (T) makes the reduced complexity of system.

Lift now M and be 64, K is 10 and K ₁Be 10 for example explains, so be not limited to this.In the present embodiment, 10 better points P (1) of the first estimation layer can be obtained by several different methods, lift now the weights enumeration method and how much enumeration methods are that example explains.Please refer to Fig. 4, it illustrates according to the real part candidate value that corresponds to the first estimation layer of the first embodiment of the invention schematic diagram to an example of imaginary part candidate value.In Fig. 4, lift M ₁Be 5 and M ₂Be 4 for example explains, so be not limited to this.In Fig. 4, for x _TFirst force zero soft output solution, obtain 5 real part candidate values and 4 imaginary part candidate values based on the SE enumeration method, and 10 better points P (1) of the first estimation layer with the weights enumeration method by obtaining in 5 real part candidate values and 4 imaginary part candidate value the corresponding a plurality of first starlike points.If x _TFirst when forcing zero soft defeated solution to be the some Z (1.4,2.3) of Fig. 2,5 real part candidate values are respectively 5 nearest value of real part of value of real part 1.4 of order apart from Z on the real part axle, are for example 1,3 ,-1,5 to reach-3.And 4 imaginary part candidate values are respectively 5 nearest imaginary values of imaginary values 2.3 of ordering apart from Z on the imaginary part axle, are for example 3,1,5 and-1.

For the weights enumeration method, x _TFirst force zero corresponding 10 the better points P of soft output solution (1) to be included in to correspond to N ₁Individual real part candidate value and N ₂In the first starlike point of individual imaginary part candidate value.That is, for { 1≤N ₁≤ M ₁, 1≤N ₂≤ M ₂The connection collection, if satisfy N ₁N ₂≤ K ₁, K ₁Individual better points can be included in { N ₁, N ₂Formed the first subclass of connection collection in.Wherein, N ₁And N ₂Be positive integer.In Fig. 4, { N ₁, N ₂The connection collection be the shadow region, 10 better points P (1) namely are included in 15 the first starlike points of this shadow region.Wherein, any the corresponding PED in these 10 better points P (1) is less than any the corresponding PED in other the first starlike point.

For instance, the first starlike point of comprising of the zone 402 of Fig. 4 is N ₁And N ₂Product less than or equal to 5 set of multiply by the first starlike point of 2 (equaling 10).Zone 404 the first starlike points that comprise are N ₁And N ₂Product less than or equal to 3 set of multiply by the first starlike point of 3 (equaling 9).Zone 406 the first starlike points that comprise are N ₁And N ₂Product less than or equal to 2 set of multiply by the first starlike point of 4 (equaling 8).The shadow region is connection collection, shadow region that is the first subclass that corresponds to three set in zone 402,404 and 406.

In addition, 10 better points P (1) of the first estimation layer can also obtain near a plurality of starlike point of how much enumeration methods first of xT forces zero soft output solution.Please refer to Fig. 5, it illustrates according to the real part candidate value that corresponds to the first estimation layer of first embodiment of the invention another routine schematic diagram to the imaginary part candidate value.In Fig. 5, lift M ₁Be 5 and M ₂Be 3 for example explains, so be not limited to this.In Fig. 5, with how much enumeration methods by obtaining 10 better points P (1) of the first estimation layer in 5 real part candidate values and 3 corresponding 15 the first starlike points of imaginary part candidate value.That is 10 better points P (1) namely are included in 15 the first starlike points of this shadow region.That is, make respectively x _TEqual these 15 starlike points, and obtain respectively r _TTx _T, and resulting result is defined as respectively y ' _TThen, these 15 the first corresponding PED of starlike point ((y ' _T-y _T) ²Square root) in minimum 10 elect better points P (1) as.10 better points P (1) of the first estimation layer be assumed to be A1, B1 ... and J1.

These 10 better points P (1) are passed to the second estimation layer, and correspond to respectively 10 the second subclass (shadow region 601～610).Please refer to Fig. 6, it illustrates according to the real part candidate value that corresponds to the second estimation layer of the first embodiment of the invention schematic diagram to an example of imaginary part candidate value.Lift now K in Fig. 6 ₂Be 9 for example explains, so be not limited to this.In Fig. 6, accept 10 the better points P (1) from the first estimation layer, that is A1, B1 ... and J1, make respectively x _TEqual A1, B1 ... and J1 substitution formula y _T-1=r _{(T-1) T}x _T+ r _{(T-1) (T-1)}x _T-1In, with try to achieve respectively correspond to A1, B1 ... and 10 x of the second estimation layer of J1 _T-1Second force zero soft output solution.Then, based on the SE enumeration method, corresponding to 10 x _T-1Second force zero soft output solution, obtain separately M ₁Individual real part candidate value and M ₂Individual imaginary part candidate value.

In Fig. 6, obtain x according to the first better points A1 that estimates layer _T-1Second force zero soft output solution after, for example obtain M based on the SE enumeration method ₁=6 real part candidate values and M ₂=5 imaginary part candidate values.Then, suppose K=12, for example obtained 21 the second starlike points of the second subclass (shadow region 601) with the weights enumeration method by 6 real part candidate values and 5 imaginary part candidate values.Then, calculate the PED of these 21 the second starlike points, and from then on take out K=12 PED reckling in 21 the second starlike points, be assumed to be A1_1, A1_2 ..., reach A1_12.Take A1_1 as example, correspond to A1_1 PED ask method as follows.Schilling x _TEqual A1, x _T-1Equal A1_1 substitution formula r _{(T-1) T}x _T+ r _{(T-1) (T-1)}x _T-1In, resulting value is defined as y ' _T-1Then, by x _TCorresponding y ' during=A1 _T(=r _TTx _T), calculate (y ' _T-y _T) ²+ (y ' _T-1-y _T-1) ²Square root.This result is the PED that corresponds to A1_1.

In like manner, the x that obtains according to the first better points B1 that estimates layer _T-1Second force zero soft output solution after, for example obtain M based on the SE enumeration method ₁=5 real part candidate values and M ₂=4 imaginary part candidate values.Suppose K=10, can obtain corresponding to 10 better points of B1, be for example B1_1, B1_2 ..., and B1_10.In like manner, the x that obtains according to the first better points J1 that estimates layer _T-1Second force zero soft output solution after, for example obtain M based on the SE enumeration method ₁=2 real part candidate values and M ₂=3 imaginary part candidate values, and obtain K=6 better solution, be for example J1_1, J1_2 ... reach J1_6.

Then, (that is be to take out K in A1_1～A1_12, B1_1～B1_10..., J1_1～J1_6) at the starlike point of this (12+10+...+6) individual second ₂The second starlike point that (=9) individual PED is less is with the K as the second estimation layer ₂(=9) individual better points P (2), and make its be respectively A2, B2 ... and I2.Then, these 9 better points P (2) are passed to the 3rd estimation layer.Wherein, any the corresponding PED in these 9 better points P (2) is less than any the corresponding PED in other the second starlike point.

During less than or equal to (T-2), the n value is added 1 when the n value.Thus, corresponding to n estimation layer, in like manner, accept the K from (n-1) estimation layer _(n-1)Individual better points P (n-1), that is A (n-1), B (n-1) ..., obtain corresponding to the K that n estimates layer _(n-1)Individual x _T-n+1N force zero soft output solution.And based on the SE enumeration method, select this K _(n-1)Individual n forces the 1 starlike point of the corresponding a plurality of n of soft output solution, and according to the corresponding PED of the starlike point of these n, obtains the K of n estimation layer _nIndividual better points P (n).Wherein, the K of (T-1) estimation layer _(T-1)Individual better points P (T-1) is passed to T estimation layer.K _(T-1)Be for example 6.

Please refer to Fig. 7, its T that corresponds to that illustrates according to first embodiment of the invention estimates the real part candidate value of layer to the schematic diagram of an example of imaginary part candidate value.In Fig. 7, accept 6 the better points P (T-1) from (T-1) estimation layer, substitution solution y ₁=r _1Tx _T+ ...+r ₁₁x ₁And obtain corresponding to 6 x of T estimation layer ₁T force zero soft output solution.Based on the SE enumeration method, for 6 x ₁T force zero soft output solution to obtain separately 1 real part candidate value and 1 imaginary part candidate value, as shown in the shadow region of Fig. 7.These real part candidate values and imaginary part candidate value correspond to 6 starlike points of T, and these 6 starlike points of T are 6 better points P (T) of T estimation layer.Correspond to minimum PED person in these 6 better points P (T), be the optimum solution that corresponds to T signal.

the disclosed sphere decoding method that is applied to the MIMO passage of first embodiment of the invention, in a plurality of territories, based on the SE enumeration method, according to the corresponding signal to noise ratio of each estimation layer, the corresponding PED of the starlike point of part that only need calculate corresponding number at each estimation layer namely can obtain one of T signal optimum solution, and need not calculate the corresponding PED of whole starlike point, significantly reduce the estimation complexity of sphere decoding method, and significantly reduce the needed time of computing, therefore can be received according to R reception antenna of receiving terminal R receives signal, the T root that solves rapidly the transmission end transmits the value of T signal of antenna output.Therefore use the receiving terminal of sphere decoding method of the MIMO passage of the present embodiment can have dynamical signal transmission and signal decoding, and reduce the hardware complexity of system and reduce the advantage of chip area.In addition, system can determine M for the MIMO channel characteristic ₁, M ₂, K and K _nThe selection of value makes system have minimum packet error probability or bit error rate, keeps the signal receiving performance of system.

The second embodiment

The sphere decoding method that is applied to multi-input multi-output channel (MIMO) of the present embodiment is applied in real number field, and it comprises: a. receives T the signal that transmits via this MIMO passage, and T is positive integer, and this MIMO passage is take an access matrix as feature; B. this access matrix is launched and produce a corresponding triangular matrix with a real number form, this triangular matrix corresponds to one first estimation layer to a 2T estimation layer, and wherein each estimation layer comprises

Individual starlike point,

Be positive integer, wherein the corresponding signal to noise ratio of this j estimation layer is that j is 1 positive integer to (2T-1) more than or equal to the corresponding signal to noise ratio of this (j+1) estimation layer; C. try to achieve and correspond to one first of this first estimation layer and force zero soft output solution, and obtain the K of this first estimation layer based on the SE enumeration method _1-Individual better points P (1), K ₁Less than or equal to D. according to this K of this first estimation layer _1-Individual better points P (1) obtains corresponding to the K of one second estimation layer ₁Individual second forces zero soft output solution, and based on the SE enumeration method, selects this K ₁Individual second forces 1 soft output solution the corresponding a plurality of second starlike point, and according to these the second corresponding part of Euclidean distance of starlike point (PED), obtains the K of the second estimation layer ₂Individual better points P (2); E. estimate this K of layer according to this (n-1) _(n-1)Individual better points P (n-1) obtains corresponding to the K that a n estimates layer _(n-1)Individual n forces zero soft output solution, and based on the SE enumeration method, selects this K _(n-1)Individual n forces the 1 starlike point of the corresponding a plurality of n of soft output solution, and according to the corresponding PED of the starlike point of these n, obtains the K of this n estimation layer _nIndividual better points P (n), n are 3～(2T-1) positive integer; And f. is according to this K of this (2T-1) estimation layer _(2T-1)Individual better points P (2T-1) obtains corresponding to the K that this 2T estimates layer _(2T-1)Individual 2T forces zero soft output solution, and based on the SE enumeration method, obtains the K of this 2T estimation layer _2TIndividual better points P (2T), and according to this K _2TCorrespond to minimum PED person in individual better points P (2T), produce an optimum solution that corresponds to this T signal.Wherein, K ₁, K ₂... and K _2TBe positive integer, and K ₁＞=K ₂＞=...＞=K _2T

Further be described in detail as follows now.Please refer to Fig. 8, it illustrates the flow chart according to the sphere decoding method that is applied to the MIMO passage of second embodiment of the invention.At first, in step S800, receive T the signal that transmits via a MIMO passage, this MIMO passage is 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.Equation with the mimo system of time domain or frequency domain representation is defined as suc as formula (1):

r＝Hx+n

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 S810, access matrix H is launched with a real number form, and the access matrix H that launches with real number form is carried out a Q-R decompose to produce a triangular matrix, this triangular matrix corresponds to one first estimation layer to a 2T estimation layer.Preferably, the first estimation layer～2T estimation layer corresponds in triangular matrix, matrix element be zero element number maximum, inferior many ..., a minimum column element.In step S810, the equation of mimo system (1) be in fact be converted to as follows:

r _R＝H _Rx _R+n _R (4)

Wherein, $r_R=\left|\begin{array}{c}\mathrm{Re}\left(r\right)\\ \mathrm{Im}\left(r\right)\end{array}\right|,$ $x_R=\left|\begin{array}{c}\mathrm{Re}\left(x\right)\\ \mathrm{Im}\left(x\right)\end{array}\right|,$ $n_R=\left|\begin{array}{c}\mathrm{Re}\left(n\right)\\ \mathrm{Im}\left(n\right)\end{array}\right|,$ $H_R=\left[\begin{array}{cc}\mathrm{Re}\left(H\right)& -\mathrm{Im}\left(H\right)\\ \mathrm{Im}\left(H\right)& \mathrm{Re}\left(H\right)\end{array}\right].$

In addition, decompose and can get by above-mentioned Q-R:

H _R＝Q _RR _R

Q wherein _RBe a 2R ' * 2T matrix, and R _RIt is the triangular matrix of one 2T * 2T.This triangular matrix R _RBe for example a upper triangular matrix, but be not limited to this.After step S810, another equation that can obtain mimo system according to equation (4) is as follows:

$y_R=Q_R^T{r}_R=R_R{x}_R+Q^T{n}_R---\left(5\right)$

Wherein, Q ^TWith the product of Q be unit matrix I.If the noise item Q with equation (5) ^Hn _RIgnore, and triangular matrix R _RWhen for example being upper triangular matrix, equation (5) can be unfolded as follows by matrix form:

(6)

Wherein,

And

Be known, p is the positive integer of 1～2T, and q is the positive integer of 1～2T.The present embodiment will be by known

And

Value, obtain

Optimum solution.

In equation (6), correspond to

Equation Be defined as the first estimation layer, correspond to

Equation Be defined as the second estimation layer, correspond to ,

Equation sequentially be defined as the 3rd estimation layer～(2T-1) and estimate layer, correspond to

Equation Be defined as 2T estimation layer.Wherein, each estimation layer comprises

Individual starlike point,

Be positive integer.The K-Best sphere decoding method produces K corresponding to n estimation layer _nIndividual better points, and K _nLess than

N=1～2T.

Then, in step S820, try to achieve and correspond to the first estimation layer

One first force zero soft output solution, and obtain the K of the first estimation layer based on the SE enumeration method ₁Individual better points P (1), K ₁Be positive integer and K ₁Less than or equal to

Wherein, corresponding to the first estimation layer, be solution

And try to achieve

First force zero soft output solution

First force the zero soft defeated solution may be with floating number.Above-mentioned SE enumeration method according to

The first distance of forcing the one dimension between zero soft defeated solution, sequentially list from small to large Near the first K that forces zero soft defeated solution ₁Individual integer value, this K ₁Individual integer value is K ₁Individual better points P (1).Wherein, this K ₁Any corresponding PED in individual better points P (1) is less than any the corresponding PED in other the first starlike point.Must not calculate in fact the K that PED can find the first estimation layer in the first estimation layer ₁Individual better points P (1) makes the reduced complexity of system.

Then, in step S830, according to the K of the first estimation layer ₁Individual better points P (1) obtains corresponding to the K of the second estimation layer ₁Individual

Second force zero soft output solution, and based on the SE enumeration method, definition K ₁Individual second forces zero soft output solution to distinguish corresponding a plurality of the second subclass, this K ₁Individual the second subclass takes out K the second starlike point separately, and wherein each second corresponding K value of subclass is identical or different.Then according to the corresponding PED of these second starlike points, obtain the K of the second estimation layer ₂Individual better points P (2).

In step S830, according to the K of the first estimation layer ₁Individual better points P (1) separates And obtain corresponding to the K of the second estimation layer ₁Individual

Second force zero soft output solution, and based on the SE enumeration method, for K ₁Individual

Second force zero soft output solution to obtain separately K the second starlike point.

Then, make respectively the value of these the second starlike points be

With its value substitution with corresponding better points P (1) And resulting result is defined as

So, the corresponding PED of each second starlike point is corresponding

Square root.Obtain the K of the second estimation layer in these second starlike points ₂Individual better points P (2), and with this K ₂Individual better points P (2) is passed to the 3rd estimation layer.Wherein, this K ₂Any corresponding PED in individual better points P (2) is less than any the corresponding PED in other the second starlike point.

Then, in step S840, according to the K of (n-1) estimation layer _(n-1)Individual better points P (n-1) obtains corresponding to the K that n estimates layer _(n-1)Individual

N force zero soft output solution, and based on the SE enumeration method, select K _(n-1)Individual n forces the 1 starlike point of the corresponding a plurality of n of soft output solution, and according to the corresponding PED of the starlike point of these n, obtains the K of n estimation layer _nIndividual better points P (n), n are 3～(2T-1) positive integer.

Step S840 arranges in fact a plurality of starlike point of n estimation layer with the SE enumeration method, and defines a plurality of n subclass of n estimation layer.In step S840, when the value of n was 3, at first foundation second was estimated the K of layer ₂The corresponding PED of individual better points P (2) sequentially arranges K from small to large ₂Individual better points P (2).Then, estimate the K of layer according to second after arranging ₂Individual better points P (2) separates

And obtain corresponding to the K of the 3rd estimation layer ₂Individual

The 3rd force zero soft output solution, and based on the SE enumeration method, for K ₂Individual

The 3rd force zero soft output solution to obtain K ₂Individual three subsetss close, and these three subsetss close and comprise separately M _iThe 3rd starlike point,, M _iLess than or equal to

Then, make respectively the value of these the 3rd starlike points be

, substitution

And resulting result is defined as

So, the corresponding PED of each the 3rd starlike point is corresponding

Square root.Obtain the K of the 3rd estimation layer in these the 3rd starlike points ₃Individual better points P (3), and with this K ₃Individual better points P (3) is passed to the 4th estimation layer.Wherein, this K ₃Any corresponding PED in individual better points P (3) is less than any the corresponding PED in other the 3rd starlike point.

During less than or equal to (2T-2), the n value is added 1 when the n value.Thus, estimate the K of layer according to (n-1) _(n-1)Individual better points P (n-1) separates

And obtain corresponding to the K of n estimation layer _(n-1)Individual

N force zero soft output solution, and based on the SE enumeration method, each

N force zero a soft output solution corresponding n subclass separately, that is correspond to M separately _iThe starlike point of individual n.From the starlike point of these n, obtain the K of n estimation layer _nIndividual better points P (n).Wherein, the corresponding PED of the starlike point of n is corresponding

Square root.Wherein, this K _nAny corresponding PED in individual better points P (n) is less than any the corresponding PED in the starlike point of other n.Thus, must not calculate nearly K in n estimation layer _(n-1)M PED can find the K of n estimation layer _nIndividual better points P (n) makes the reduced complexity of system.

Afterwards, in step S850, according to the K of (2T-1) estimation layer _(2T-1)Individual better points P (2T-1) separates

And obtain corresponding to the K of 2T estimation layer _(2T-1)Individual 2T force zero soft output solution, and based on the SE enumeration method, obtain the K of 2T estimation layer _2TIndividual better points P (2T), and according to K _2TCorrespond to minimum PED person in individual better points P (2T), produce and correspond to one of T signal optimum solution, that is be matrix x _ROptimum solution.Wherein, this K _2TIn individual better points P (2T), each self-corresponding part of Euclidean distance is corresponding

Square root.Thus, must not calculate nearly in 2T estimation layer

x K _(2T-1)Inferior PED can find the K of 2T estimation layer _2TIndividual better points P (T) makes the reduced complexity of system.

Lift now M and be 64, K ₁Be 5 and K ₂Be 10 for example explains, so be not limited to this.Corresponding to

First force zero soft defeated solution, based on the SE enumeration method, must not calculate the K that PED can obtain the first estimation layer ₁=5 better points P (1), and these 5 better points P (1) are passed to the second estimation layer.Any corresponding PED in these 5 better points P (1) is less than any the corresponding PED in other the first starlike point.

In the present embodiment, 10 better points P (2) of the second estimation layer can be obtained by several different methods, lift now the weights enumeration method and how much enumeration methods are that example explains.Please refer to Fig. 9, it illustrates the schematic diagram to an example of the second starlike point according to the better points P (1) that corresponds to the second estimation layer of second embodiment of the invention.In Fig. 9, for K ₁=5 Second force zero soft output solution, obtain 5 the second subclass based on the SE enumeration method along a single axle, each second subclass obtains M separately ₂The individual second starlike point, and 10 better points P (2) of the second estimation layer with the weights enumeration method by obtaining in a plurality of second starlike point.Wherein, each second forces zero soft output solution corresponding M separately ₂Value, each second forces zero soft output to separate each self-corresponding M ₂Value be identical or different.In Fig. 9, lift the M of the 1st～5 row ₂Be respectively 4,4,3,2 and 2 for example explains, so be not limited to this.

Wherein, due to R _RBe real number matrix, R _RAll elements be real number, therefore, obtaining a plurality of second starlike point based on the SE enumeration method along single axle and refer to obtain a plurality of second starlike point based on the SE enumeration method along one of them of real part axle or imaginary part axle herein.In Fig. 9, the starlike point of each row second corresponds to respectively a better points P (1), and is sequentially increased by the most above-listed extremely corresponding PED value of the most following corresponding better points P (1).That is to say, the second starlike point of first row shadow region is, based on the minimum better points P (1) of PED value, and second value of forcing near the real part of four the second starlike points zero soft output solution of trying to achieve, or the value of the imaginary part of four the second starlike points.And the second starlike point of secondary series shadow region is, based on the little better points P of PED value second (1), and second value of forcing near the real part of four the second starlike points zero soft output solution of trying to achieve, or the value of the imaginary part of four the second starlike points.Remaining by that analogy.

For the weights enumeration method, Second force zero corresponding 10 the better points P of soft output solution (2) to comprise to correspond to the N of number from top to bottom ₁Individual better points P (1) by from left to right the number N ₂The individual second starlike point.That is, for { 1≤N ₁≤ M ₁, 1≤N ₂≤ M ₂The connection collection, if satisfy N ₁N ₂≤ K ₂, K ₂Individual better points can be included in { N ₁, N ₂Formed the second subclass of connection collection in.Wherein, N ₁And N ₂Be positive integer.In Fig. 9, { N ₁, N ₂The connection collection be the shadow region, 10 better points P (2) namely are included in 15 the second starlike points of this shadow region.Wherein, any the corresponding PED in these 10 better points P (2) is less than any the corresponding PED in other the second starlike point.

In addition, 10 better points P (2) of the second estimation layer can how much enumeration method from

Near the second a plurality of starlike point of forcing zero soft output solution in obtain.Please refer to Figure 10, it illustrates another routine schematic diagram to the second starlike point according to the better points P (1) that corresponds to the second estimation layer of second embodiment of the invention.In Figure 10, lift M ₂Be 3 for example explains, so be not limited to this.In Figure 10, to obtain 10 better points P (2) of the second estimation layer in 15 second starlike points of how much enumeration methods by the shadow region.These 10 better points P (2) are passed to the 3rd estimation layer.

Please refer to Figure 11, it illustrates the schematic diagram to an example of the 3rd starlike point according to the better points P (2) after the arrangement that corresponds to the 3rd estimation layer of second embodiment of the invention.In Figure 11, accept 10 the better points P (2) from the second estimation layer, and sequentially arrange 10 better points P (2) according to the size of the corresponding part of Euclidean distance of 10 better points P (2).Then, according to 10 the better points P (2) after arranging, substitution solution respectively

And obtain corresponding to 10 of the 3rd estimation layer

The 3rd force zero soft output solution.Then, based on the SE enumeration method, for 10 The 3rd force zero soft output solution to obtain separately M ₃The individual the 3rd starlike point.

In Figure 11, with the weights enumeration method by 10

The 3rd force 9 the better points P (3) that obtain the 3rd estimation layer in zero soft output solution the corresponding a plurality of the 3rd starlike point, so be not limited to this, can also obtain by how much enumeration methods.In Figure 11, each the 3rd forces the zero corresponding M of soft output solution ₃The starlike point of individual n comprises and corresponds to V ₁The V of individual better points P (2) ₂The individual the 3rd starlike point.That is, for { 1≤V ₁≤ K ₃, 1≤V ₂≤ M ₃The connection collection, if satisfy V ₁V ₂≤ K ₃, K ₃Individual better points can be included in { V ₁, V ₂Connection collection formed three subsetss close.Wherein, V ₁And V ₂Be positive integer.In Figure 11, { V ₁, V ₂The connection collection be the shadow region, 9 better points P (3) namely are included in 20 the 3rd starlike points of this shadow region.Wherein, any the corresponding PED in these 9 better points P (3) is less than any the corresponding PED in other the 3rd starlike point.

For instance, the 3rd starlike point that comprises of the zone 1102 of Figure 11 is V ₁And V ₂Product less than or equal to 10 set of multiply by the 3rd starlike point of 1 (equaling 10).Zone 1104 the 3rd starlike points that comprise are V ₁And V ₂Product less than or equal to 5 set of multiply by the 3rd starlike point of 2 (equaling 10).Zone 1106 the 3rd starlike points that comprise are V ₁And V ₂Product less than or equal to 3 set of multiply by the 3rd starlike point of 3 (equaling 9).Zone 1108 the 3rd starlike points that comprise are V ₁And V ₂Product less than or equal to 2 set of multiply by the 3rd starlike point of 5 (equaling 10).Shadow region (three subsetss close) is the connection collection that corresponds to four set in zone 1102,1104,1106 and 1108.

During less than or equal to (2T-2), the n value is added 1 when the n value.Thus, corresponding to n estimation layer, in like manner, accept the K from (n-1) estimation layer _(n-1)Individual better points P (n-1) obtains corresponding to the K that n estimates layer _(n-1)Individual

N force zero soft output solution, and based on the SE enumeration method, select this K _(n-1)Individual n forces the 1 starlike point of the corresponding a plurality of n of soft output solution, and according to the corresponding PED of the starlike point of these n, obtains the K of n estimation layer _nIndividual better points P (n).Wherein, the K of (2T-1) estimation layer _nIndividual better points P (2T-1) is passed to 2T estimation layer.

Please refer to Figure 12, it illustrates the schematic diagram to an example of the starlike point of 2T according to the better points P (2T-1) after the arrangement that corresponds to 2T estimation layer of second embodiment of the invention.In Figure 12, accept 6 the better points P (2T-1) from (2T-1) estimation layer, the substitution solution And obtain corresponding to 6 of 2T estimation layer

2T force zero soft output solution.Based on the SE enumeration method, for 6

2T force zero soft output solution to obtain separately 1 starlike point of 2T, as shown in the shadow region of Figure 12.These 6 starlike points of 2T are 6 better points P (2T) of 2T estimation layer.Correspond to minimum PED person in these 6 better points P (2T), be the optimum solution that corresponds to T signal.

The disclosed sphere decoding method that is applied to the MIMO passage of second embodiment of the invention, in real number field, based on the SE enumeration method, only need the corresponding PED of the starlike point of calculating section namely can obtain the optimum solution of T signal, and need not calculate the corresponding PED of whole starlike point.So, can significantly reduce the estimation complexity of sphere decoding method, and significantly reduce the needed time of computing, therefore can be received according to R reception antenna of receiving terminal R receives signal, the T root that solves rapidly the transmission end transmits the value of T signal of antenna output.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, and reduces the advantage of hardware complexity and the minimizing chip area of system.In addition, system can determine K for the MIMO channel characteristic ₁～K _2T, K, M _iThe selection of value, make system have minimum packet error probability or bit error rate, keep the signal receiving performance of system.

The disclosed sphere decoding method that is applied to multi-input multi-output channel of the above embodiment of the present invention, in a plurality of territories and real number field, based on the SE enumeration method, and be able to from a plurality of starlike point of each estimation layer, select the starlike point of part to calculate its corresponding PED, can obtain K better points of each estimation layer.Thus, not only significantly reduce the estimation complexity of sphere decoding method, and significantly reduce the needed time of computing, and can keep the signal receiving performance of system.

In sum, although the present invention with a preferred embodiment openly as above, so it is not to limit the present invention.Those skilled in the art of 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 (8)

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

T the signal that reception transmits via the MIMO passage, T is positive integer, the MIMO passage is take an access matrix as feature;

Correspond to a n estimation layer according to the MIMO access matrix, arrange a plurality of starlike point of n estimation layer with an enumeration method, and define at least one n subclass of n estimation layer, this n subclass has a plurality of starlike point of this n estimation layer, n is positive integer, wherein the MIMO access matrix corresponds to the 1st～the T estimation layer, n estimation layer and K _nCorrespond to respectively the 2nd～the (T-1) estimation layer and K ₂～K _(T-1)One of them, Kn is positive integer;

Take out respectively K individual starlike as better points from each n subclass, and selected the K of n estimation layer by the better points of all n subclass _nIndividual better points, K is positive integer, K _nBe the positive integer less than or equal to the better points number summation of all n subclass;

Transmit K by the 1st estimation layer ₁Individual better points to the 2 estimation layers, the 2nd estimation layer transmits K ₂Individual better points to the 3 estimation layers ..., (T-1) estimation layer transmits K _(T-1)Individual better points to the T estimation layer; And

K according to T estimation layer _TIndividual better points determines an optimum solution;

Wherein, K and K ₁～K _TThe number size one of them is determined by MIMO access matrix characteristic at least, T wherein, K ₁, K ₂, K _(T-1), K _TBe positive integer.

2. the method for claim 1, wherein the 1st, 2 ..., after T estimation layer corresponded to respectively the MIMO access matrix and decomposes via Q-R, zero element was maximum, inferior many ..., minimum row, wherein K ₁〉=K ₂〉=... 〉=K _T

3. the method for claim 1, wherein MIMO access matrix characteristic means that the MIMO access matrix corresponds to the reliability of forcing zero soft output solution of each estimation layer.

4. method as claimed in claim 3, wherein the reliability of each estimation layer depends on the signal to noise ratio intensity of each estimation layer.

5. method as claimed in claim 3 is if wherein the reliability of a estimation layer is greater than the reliability of b estimation layer K _aGreater than K _b, wherein a and b estimation layer is respectively wherein two estimation layers of the 1st～T estimation layer, and a estimation layer is K _aIndividual better points, b estimation layer is K _bIndividual better points, K _a, K _bBe positive integer.

6. the method for claim 1, wherein this enumeration method is SE (Schnorr﹠amp; Euchner) enumeration method.

7. the method for claim 1, wherein take out K starlike step as better points and take out the starlike as better points of K minimum PED according to part of Euclidean distance PED from the n subclass from the n subclass.

8. method as claimed in claim 7, wherein this optimum solution is the K of this T estimation layer _TPED value reckling in the middle of individual better points.

Images