CN102136889A - Bit-level PDA detection method and device of high-order QAM multiaerial system - Google Patents

Abstract

The invention discloses an auxiliary bit-level probability data detection method and device of a high-order QAM (quadrature amplitude modulation) multiaerial system. The method comprises the steps of: modulating the bit flow at a sending end into a modulating symbol; distributing the modulating symbol to all sending antennas, and then sending the modulating symbol to a receiving end; and receiving the modulating symbol by the receiving end, analyzing the bit flow according to the modulating symbol, and detecting the probability data on the basis of the bit flow. Compared with the existing algorithm, the bit-level PDA detection method and device reduce the complexity and solves the detection problem of data of all layers in a high-order modulated MIMO (multiple input multiple output) system.

Description

Bit-level PDA detection method and device at the high-order QAM multiaerial system

Technical field

The invention belongs to communication technical field, specifically, belong to the detection range in the mimo system.Relate in particular at high-order QAM (quadrature amplitude modulation) multiaerial system (Multiple-Input Multiple-Output, MIMO) bit-level probability data aided detection method and device.

Background technology

Mimo system is can simple defining as follows: given any one wireless communication system, consider such link, and its transmitting terminal and receiving terminal use many antennas simultaneously.Mimo system idea behind is to make that the signal between many antennas of transmitting terminal and the many antennas of receiving terminal is that " combination " forms.From studies show that of information theory view, use the multi-input multi-output system of many antennas can increase the capacity of wireless communication system simultaneously at transmitting terminal and receiving terminal, its channel capacity is directly proportional with the minimum value of dual-mode antenna number, and can improve the performance of system greatly.When the QAM modulation of using high-order, can obtain the good compromise of transmission rate and bit error rate especially, be fit to very much the requirement of B3G, 4G wireless communication system high speed business.The shortcoming that mimo system is compared point-to-point communication is that the receiver complexity is higher, especially high level MIMO system.Therefore, become in the next generation wireless communication field in the hot technology, be necessary very much to study low complex degree detection algorithm at high level MIMO system in the MIMO technology.

Probability data auxiliary (PDA) detects and is used in target tracking domain at first, all is widely used in a lot of fields subsequently.The PDA algorithm is generalized in the Multiuser Detection of CDMA and mimo system gradually.In these fields, the PDA algorithm is normal and combine based on decoder, detector, the equalizer of maximum a posteriori probability, relies on it to soft information processing, changes the complexity that has subtracted system.Its main thought is: according to central-limit theorem with some variablees and be approximately gaussian variable.PDA detects the current detection symbol is used as stochastic variable, other layers interference is approximated to Gaussian noise, by this approximate, prior information according to other layers symbol is calculated the average and the variance of mating with gaussian variable, and then obtain the conditional probability of current detection symbol, this conditional probability is again as the prior information that detects other layers symbol, the conditional probability of each layer symbol is able to continuous iteration renewal like this, if certain detection signal log-likelihood ratio that twice iteration obtains illustrates then that less than a set point detection of this signal restrains.

Existing P DA algorithm basic procedure:

Received signal r=Hx+n (1)

Wherein, r is a received signal, and x sends signal phasor, and H is a channel matrix, and n is the white Gaussian noise vector.

If m the symbol x of x _mBe current detection symbol, h _kRepresent the k row of channel matrix, then

$r=h_m{x}_m+\underset{k\≠m}{\mathrm{\Σ}}{h}_k{x}_k+n---\left(2\right)$

Set $w=\underset{k\≠m}{\mathrm{\Σ}}{h}_k{x}_k+n---\left(3\right)$

W is by a gaussian variable n and several stochastic variables h _kx _kAddition forms.According to central-limit theorem, work as h _kx _kNumber when abundant, w trends towards Gaussian Profile.Average μ _wWith covariance matrix R _wFor:

${\mathrm{\μ}}_w=E\{\underset{k\≠m}{\mathrm{\Σ}}{h}_k{x}_k+n\}=\underset{k\≠m}{\mathrm{\Σ}}E\left\{x_k\right\}{h}_k$

$R_w=E\left\{(w-{\mathrm{\μ}}_w)(w-{\mathrm{\μ}}_w{)}^H\right\}=\underset{k\≠m}{\mathrm{\Σ}}\mathrm{Var}\left\{x_k\right\}{h}_k{h_k}^H+{\mathrm{\σ}}_n^{2}I$

Variance for noise.I is a unit matrix.E{} represents to ask expectation, and Var{} represents to ask variance.

Each component of supposing w all is the circulation symmetry, and promptly the real part of random vector is identical with variance with distribution and average with imaginary part, can obtain the current detection symbol and get a _jPosterior probability be:

${\mathrm{Pr}}_m(x_m=a_j|r)=\frac{p\left(r\right|x_m=a_j)\mathrm{Pr}(x_m=a_j)}{p\left(r\right)}\mathrm{cexp}(-(r-a_j{h}_m-{\mathrm{\μ}}_w{)}^H{R}_w^{-1}(r-a_j{h}_m-{\mathrm{\μ}}_w))---\left(4\right)$

A wherein _j∈ Q, Q are the modulation symbol set.

Can find Pr by above-mentioned flow process _m(x _m=a _j| value r) will depend on μ _wAnd R _w, and this two values and x _k, the probability distribution of k ≠ m is relevant.In first round iteration, x during m=1 _k, the distribution of k ≠ m is initialized as equiprobability and distributes.By the calculating of (4), can obtain about x ₁A new distribution.When m=2, the E{x that relates in (4) ₁, Var{x ₁All calculate according to new distribution.Up to the distributed update of all data one time, first round iteration finishes by that analogy.Thereafter the distribution of central all symbols of the iteration of a new round is all according to the distribution after upgrading.Up to convergence.

Basic procedure by above-mentioned existing P DA algorithm can draw, and the core of PDA algorithm is the Gaussian approximation to noise and interlayer interference, this approximate finding the solution of simplifying posterior probability.But have research to point out that the accuracy of Gaussian approximation has material impact to the performance of PDA, if number of users or number of transmit antennas are less, central-limit theorem will no longer be set up, and it is nonsensical that Gaussian approximation just becomes, and the performance of PDA will variation.

In addition, PDA depends on the conditional likelihood likelihood ratio to the demodulation of bit

$L\left(c_k\right)=\log \frac{{\mathrm{\Σ}}_{a_j\∈Q^{+}}\mathrm{Pr}\left(a_j\right|r)}{{\mathrm{\Σ}}_{a_j\∈Q^{-}}\mathrm{Pr}\left(a_j\right|r)}---\left(5\right)$

Wherein:

Q ⁺：＝{a _j∶c _k＝bit(a _j)＝+1}

Q ^-：＝{a _j∶c _k＝bit(a _j)＝-1}

When order of modulation higher, the number of Q institute containing element | when Q| was big, the calculative posterior probability of formula (5) was more, because the posterior probability that PDA uses is approximate posterior probability, the result of calculation of formula (5) will be very inaccurate in this case.Thereby make degradation.

To sum up, PDA is applicable to the system of multiple transmit antennas low modulation exponent number.

The posterior probability formula of using in the PDA detection (4) is to suppose that each component of w all is that the statistical property of w is by average μ at this moment under the symmetrical prerequisite of circulation _w, covariance matrix R _wDefinite fully.This sets up when initial each symbol equiprobability of hypothesis occurs, but along with the carrying out of iterative detection, the probability that each symbol occurs constantly is updated, and each component of w just no longer guarantees it is that circulation is symmetrical, and the statistical property of w is by average μ in this case _w, covariance matrix R _w, pseudo-covariance matrix K _w=E{ (w-μ _w) (w-μ _w) ^TCommon definite.Therefore, formula (4) is not accurate enough.

Based on above-mentioned cognition, Pham etc. propose broad sense PDA (GPDA), are the real signal model of equivalence with the complex signal model conversation.

$\tilde{r}=\tilde{H}\tilde{x}+\tilde{n}$

Wherein:

$\tilde{r}={\left[R\right\{r^T\left\}I\right\{r^T\left\}\right]}^T$

$\tilde{x}={\left[R\right\{x^T\left\}I\right\{x^T\left\}\right]}^T$

$\tilde{n}={\left[R\right\{n^T\left\}I\right\{n^T\left\}\right]}^T$

$\tilde{H}=\left[\begin{array}{cc}R\left\{H\right\}& -I\left\{H\right\}\\ I\left\{H\right\}& R\left\{H\right\}\end{array}\right]$

R{}, I{} represent respectively and get real part and imaginary part.Therefore the scale of signal Candidate Set dwindles, real empty two asymmetric problems of statistical property of w in the time of need not considering iteration simultaneously.But therefore the scale that changes into channel matrix behind the real signal model can enlarge twice.The PDA detection algorithm bit error rate of GPDA before comparing is lower, detects and also can reduce receiving number of antennas more traditional PDA of complexity when sending out number of antennas.

People such as Y.Jia propose CPDA, promptly still consider the complex signal model, but remove to portray the Gauss's vector that is similar to average, covariance matrix, pseudo-covariance matrix.The calculating of posterior probability does not also re-use formula (4).But use following calculation process:

V＝r-a _jh _m-μ _w

$\mathrm{\φ}\left(a_j\right)=\exp \left(\begin{array}{c}{-\left\{\begin{array}{c}R\left\{v\right\}\\ I\left\{v\right\}\end{array}\right\}{\mathrm{\Λ}}_j\left\{\begin{array}{c}R\left\{v\right\}\\ I\left\{v\right\}\end{array}\right\}}^T\end{array}\right)---\left(6\right)$

K _w＝E{(w-μ _w)(w-μ _w) ^T}

${\mathrm{\Λ}}_j={\left(\begin{array}{cc}R\{R_w+K_w\}& -I\{R_w-K_w\}\\ I\{R_w+K_w\}& R\{R_w-K_w\}\end{array}\right)}^{-1}---\left(7\right)$

$\mathrm{Pr}(x_m=a_j|r)=\frac{\mathrm{\φ}\left(a_j\right)}{\underset{a_j\∈Q}{\mathrm{\Σ}}\mathrm{\φ}\left(a_j\right)}$

CPDA be better than on the performance GPDA and before algorithm.

Summary of the invention

The present invention proposes at the bit-level probability data aided detection method (BPDA) of high-order QAM multiaerial system and device, in the mimo system of high order modulation, compares existing algorithm and has reduced complexity, has solved the detection problem of each layer data.

In order to solve the problems of the technologies described above, the present invention proposes the bit-level probability data aided detection method at the high-order QAM multiaerial system, may further comprise the steps:

The bit stream of transmitting terminal is modulated into modulation symbol;

Modulation symbol is diverted on each transmitting antenna, sends to receiving terminal;

Receive modulation symbol by receiving terminal, parse bit stream according to modulation symbol, and carry out the probability data auxiliary detection based on bit stream.

The present invention also proposes the bit-level probability data auxiliary detection device at the high-order QAM multiaerial system, comprising:

Transmitting terminal is modulated into modulation symbol with bit stream; Modulation symbol is diverted on each transmitting antenna, sends to receiving terminal;

Receiving terminal receives modulation symbol, parses bit stream according to modulation symbol, and carries out the probability data auxiliary detection based on bit stream.

The present invention also proposes the receiving terminal at the bit-level probability data auxiliary detection of high-order QAM multiaerial system, comprising:

Receiver module is used to receive modulation symbol;

Parsing module is coupled in receiver module, is used for parsing bit stream according to modulation symbol;

Detection module is coupled in parsing module, is used for carrying out the probability data auxiliary detection based on bit stream.

Traditional PDA obtains symbol and is demodulated to bit again at symbol.The present invention changes into the symbol level detection problem of high level MIMO system the bit-level detection problem of BPSK (binary phase shift keying) modulation.System modulation exponent number after the conversion is low, and the number of interlayer interference variable is more, adopts Gaussian approximation more accurate, and the relatively more suitable PDA of employing detects.Compare existing algorithm, the present invention has reduced the complexity of detection algorithm, has solved the detection problem of each layer data.

Description of drawings

Fig. 1 is the mapping naturally of 16QAM in the prior art.

Fig. 2 is the rule of the gray mappings of 16QAM in the prior art.

Fig. 3 is a 64QAM gray mappings in the prior art.

Fig. 4 is in the embodiment of the invention, the complexity of each detector contrast when signal to noise ratio is 40db, and the complexity of B-PDA is starkly lower than CPDA.

Fig. 5 is in the embodiment of the invention, the comparison of each detector false bit rate under the 16QAM situation.

Fig. 6 is in the embodiment of the invention, the comparison of each detector false bit rate under the 64QAM situation.

Fig. 7 is the bit-level probability data aided detection method flow chart that the present invention is directed to the high-order QAM multiaerial system.

Fig. 8 is the bit-level probability data auxiliary detection device structural representation that the present invention is directed to the high-order QAM multiaerial system.

Embodiment

The present invention proposes the bit-level probability data aided detection method at the high-order QAM multiaerial system.As shown in Figure 7, may further comprise the steps:

In step 701, the bit stream of transmitting terminal is modulated into modulation symbol;

In step 702, modulation symbol is diverted on each transmitting antenna, send to receiving terminal;

In step 703, receive modulation symbol by receiving terminal, parse bit stream according to modulation symbol, and carry out the probability data auxiliary detection based on bit stream.

In the step 703 of the present invention, parse according to modulation symbol in the operation of bit stream, when being linear relationship between modulation symbol and the bit stream, modulation symbol is that bit stream and generator matrix multiply each other, and calculates bit stream according to modulation symbol and generator matrix.As one embodiment of the present of invention, this generator matrix is by order of modulation, modulation system and number of transmit antennas decision.When being non-linear relation between modulation symbol and the bit stream, determine corresponding generator according to modulation symbol, determine corresponding bit stream according to generator again.

Figure 8 shows that bit-level probability data auxiliary detection device, comprising at the high-order QAM multiaerial system:

Transmitting terminal is modulated into modulation symbol with bit stream; Modulation symbol is diverted on each transmitting antenna, sends to receiving terminal;

Receiving terminal receives modulation symbol, parses bit stream according to modulation symbol, and carries out the probability data auxiliary detection based on bit stream.

Again as shown in Figure 8, the structure of receiving terminal comprises:

Receiver module is used to receive modulation symbol;

Parsing module is coupled in receiver module, is used for parsing bit stream according to modulation symbol;

Detection module is coupled in parsing module, is used for carrying out the probability data auxiliary detection based on bit stream.

As one embodiment of the present of invention, parsing module parses bit stream according to modulation symbol:

For shining upon naturally of mapping, the 64QAM modulation system naturally of 4QAM modulation system, 16QAM modulation system, it between modulation symbol and the bit stream linear relationship, described modulation symbol is that bit stream and generator matrix multiply each other, and parsing module calculates bit stream according to modulation symbol and generator matrix;

For the gray mappings of 16QAM modulation system, the gray mappings of 64QAM modulation system, between modulation symbol and the bit stream non-linear relation, parsing module is determined corresponding generator according to modulation symbol, determines corresponding bit stream according to generator again.

Because the present invention can parse bit stream according to modulation symbol, and carry out the probability data auxiliary detection based on bit stream, thereby will change into the detection based on bit stream based on the detection of modulation symbol, the high-order MIMO system has just changed into a BPSK modulated M IMO system.Mimo system behind the abbreviation is done probability data auxiliary (PDA) detect, detecting at the PDA of bit stream is to adopt existing algorithm fully.Because the attribute of PDA detection algorithm itself, PDA is more suitable for being used in the system of low-order-modulated, multiple transmit antennas.Finally, the bit stream of transmitting terminal information source data is detected by the PDA algorithm.

Be elaborated below in conjunction with accompanying drawing and embodiment.

At first, under the different modulating mode, the transfer process between bit stream and the modulation symbol describes.The present invention is primarily aimed at the modulation system of 4QAM, 16QAM, 64QAM.

1.4QAM

The modulation symbol set Q={-1-i of known 4QAM ,-1+i ,+1+i ,+1-i}.Modulation traditionally is with bit (1 ,-1), and (1 ,+1), (+1 ,+1), (+1 ,-1) are mapped to symbol in the modulation symbol set according to the modulating rule of correspondence respectively.The gray mappings rule is as shown in the table:

Table 1

Former bit	Switch bit (b)	Modulation symbol (s)
			00	-1-1	-1-i
01	-1+1	-1+i
			11	+1+1	+1+i
10	+1-1	+1-i

From former bit to switch bit is to become-1,1 with 0 to be still 1, and they shine upon one by one.So switch bit and former bit are of equal value, hereinafter no longer distinguish.It is the needs of Linear Mapping that this conversion comes from modulation conversion.For the 4QAM gray mappings is linear, so do not distinguish gray mappings or mapping naturally generally speaking.Modulation is two mappings between set, but based on needs of the present invention, this mapping one by one is converted into concrete functional relation.

Suppose from b to S, to exist linear transforming relationship

S＝bg _4QAM ^T (8)

B is 1 * log ₂The bit vectors of M.M＝|Q|。

Since at be high-order rectangle constellation, So problem equivalent in

$S_{\mathrm{real}}=b_{[1,...\frac{{\log }_{2}M}{2}]}{x}^T---\left(9\right)$

Concerning 4QAM modulation, the real part of each 4QAM symbol and imaginary part be by two independently bit controls, and can find out that the value of real part and imaginary part takes from same numerical value set from planisphere.So g _4QAMForm can be defined as g _4QAM=[x, ix], x are called the generator of 4QAM modulation.

As one embodiment of the present of invention, the real part of the S in the table 1 and first bit of b are distinguished substitution, each component of finding the solution x can obtain existing such x to satisfy (9)

x＝[1]

g _4QAM＝[1，i]

Can get thus, the gray mappings symbol of a 4QAM can be by the vectorial premultiplication g of the row of a dibit _4QAMTransposition represent.This operation is expanded to the conversion that bit under the multiaerial system flows to symbolic vector, be follow-up expression, the bit stream that this sight will be modulated is down taken advantage of and is become premultiplication with column vector B representative, the corresponding right side.Through after the conversion, a string bit stream B is modulated and shunt the process that forms the modulation symbol vector S can be expressed as:

S＝W _4QAMB

Wherein

${\mathrm{<figure-callout\; id="4"\; label="W"\; filenames="HDA0000052313360000011.png,HDA0000052313360000061.png"\; state="\{\{state\}\}">W</figure-callout>}}_{4\mathrm{QAM}}={\left(\begin{array}{ccccc}{\mathrm{<figure-callout\; id="4"\; label="g"\; filenames="HDA0000052313360000011.png,HDA0000052313360000061.png"\; state="\{\{state\}\}">g</figure-callout>}}_{4\mathrm{QAM}}& .& .& .& 0\\ .& .& & & .\\ .& & .& & .\\ .& & & .& .\\ 0& .& .& .& {\mathrm{<figure-callout\; id="4"\; label="g"\; filenames="HDA0000052313360000011.png,HDA0000052313360000061.png"\; state="\{\{state\}\}">g</figure-callout>}}_{4\mathrm{QAM}}\end{array}\right)}_{N_T\&times;2N_T}$

N _TIt is the number of transmit antennas of system.

2.16QAM

Q in 16QAM _Real=Q _Imag=3 ,-1 ,+1 ,+3},

From (1 ,-1), (1 ,+1), (+1 ,-1), (+1 ,+1), get.The mapping (natural mapping) naturally of 16QAM as shown in Figure 1.

At natural mapping, respectively the real part of modulation symbol and the front two of bit vectors are brought in the formula (9).If the real part of all modulation symbols and the front two of bit vectors all are brought into respectively in (9), the equation group that finally obtains is an overdetermination, and 4 equation group are promptly arranged, 2 unknown numbers.Separate this overdetermined equation group, obtain separating x=[2,1 under the mapping naturally], g _16QAM=[2,1,2i, i] can obtain the generator matrix W that multiaerial system 16QAM modulates naturally thus _16QAM

${\mathrm{<figure-callout\; id="16"\; label="W"\; filenames="HDA0000052313360000051.png,HDA0000052313360000061.png"\; state="\{\{state\}\}">W</figure-callout>}}_{16\mathrm{QAM}}={\left(\begin{array}{ccccc}{\mathrm{<figure-callout\; id="16"\; label="g"\; filenames="HDA0000052313360000051.png,HDA0000052313360000061.png"\; state="\{\{state\}\}">g</figure-callout>}}_{16\mathrm{QAM}}& .& .& .& 0\\ .& .& & & .\\ .& & .& & .\\ .& & & .& .\\ 0& .& .& .& {\mathrm{<figure-callout\; id="16"\; label="g"\; filenames="HDA0000052313360000051.png,HDA0000052313360000061.png"\; state="\{\{state\}\}">g</figure-callout>}}_{16\mathrm{QAM}}\end{array}\right)}_{N_T\&times;4N_T}$

N _TIt is the number of transmit antennas of system.

Can observe nature mapping and gray mappings and inequality.The rule of gray mappings as shown in Figure 2.Naturally the selection of mapping and gray mappings is to be determined by concrete signal intelligence.The high-order gray mappings is different from other mappings and is that his is non-linear.

Same hypothesis gray mappings can change into linear transformational relation such as formula (9).If x=[a, b],

Possible value set is for { 1 ,-1;-1 ,+1; + 1 ,+1; + 1 ,-1} is updated to respectively in the formula (9), gets equation group

$\left\{\begin{array}{c}-a-b=-3\\ -a+b=-1\\ a+b=+1\\ a-b=+3\end{array}\right.$

This equation group does not have and separates, and promptly gray mappings can not convert the form for the treatment of that modulation bit stream and generator matrix multiply each other to.But the combination for the part equation has

$\left\{\begin{array}{c}-a-b=-3\\ -a+b=-1\end{array}\right.\&DoubleRightArrow;a=2,b=1,$

$\left\{\begin{array}{c}a+b=+1\\ a-b=+3\end{array}\right.\&DoubleRightArrow;a=2,b=-1.$

Promptly work as

First bit be-1 o'clock, x is [2,1].When First bit be+1 o'clock, x is [2 ,-1].We can say that gray mappings is local linear, the generator partial dependency is treated modulation bit.The transforming relationship that the 16QAM gray mappings is complete is as shown in table 2.

Table 2

Modulating rule is that communicating pair is made an appointment, and has had modulating rule just can obtain the content of table 2.Calculating is stored in the receiver well before the communication beginning, and receiver is slightly to declare earlier, and the quadrant at emission symbol place determines that quadrant just can be according to the definite generator of table 2.Determined that each generator just can obtain formula (10), and the later work of expansion formula (10).

Though there is not unified linear transforming relationship in the gray mappings of 16QAM, can find by table 2, the modulation symbol of same quadrant is generated by same generator, has determined that promptly the quadrant at modulation symbol place just equals to have determined corresponding generator.The empty two parts set of the reality of QAM modulation is identical, and the generator of imaginary part is identical with the generator of real part.

3.64QAM

Q in 64QAM _Red=7 ,-5 ,-3 ,-1 ,+1 ,+3 ,+5 ,+7}, Possible value set is for { 1 ,-1 ,-1;-1 ,-1 ,+1;-1 ,+1 ,+1;-1 ,+1 ,-1; + 1 ,+1 ,-1; + 1 ,+1 ,+1; + 1 ,-1 ,+1; + 1 ,-1 ,-1; .The conversion rule of 64QAM and 16QAM is very similar, and in the natural mapping mode of 64QAM, modulation symbol can be expressed as generator matrix equally and treat the form that modulation bit stream multiplies each other.Respectively the real part of modulation symbol and the front three of bit vectors are brought in the formula (9).Obtain under shining upon naturally g _64QAM=[4,2,1,4i, 2i, 1i] can obtain the generator matrix that multiaerial system 64QAM modulates naturally thus:

$W_{64\mathrm{QAM}}={\left(\begin{array}{ccccc}{g}_{64\mathrm{QAM}}& .& .& .& 0\\ .& .& & & .\\ .& & .& & .\\ .& & & .& .\\ 0& .& .& .& g_{64\mathrm{QAM}}\end{array}\right)}_{N_T\&times;6N_T}$

Wherein, S represents modulation symbol, and B represents bit stream, and W represents generator matrix, N _TIt is the number of transmit antennas of system.

There is not unified linear transforming relationship in the gray mappings of 64QAM, but neither not have rule to follow, will

The substitution respectively of the real part of possible value and 64QAM gray mappings symbol obtains an overdetermined equation group, and this equation group does not have and separates, but the equation group that the part equation is formed has

$\left\{\begin{array}{c}-a-b-c=-7\\ -a-b+c=-5\end{array}\right.\&DoubleRightArrow;a=4,b=2,c=1$

$\left\{\begin{array}{c}+a+b+c=-3\\ -a+b-c=-1\end{array}\right.\&DoubleRightArrow;a=4,b=2,c=-1$

$\left\{\begin{array}{c}a+b-c=1\\ a+b+c=3\end{array}\right.\&DoubleRightArrow;a=4,b=-2,c=1$

$\left\{\begin{array}{c}a-b+c=5\\ a-b-c=7\end{array}\right.\&DoubleRightArrow;a=4,b=-2,c=-1$

Gray mappings with 16QAM is identical, and the generator of 64QAM gray mappings also partial dependency is treated modulation bit stream.

Table 3

As shown in Figure 3, the generator of the constellation point correspondence in same frame of broken lines is identical, have 16 different generators, the planisphere plane is divided into 16 zones, as long as determined the zone (frame of broken lines shown in Figure 3) on planisphere plane, modulation symbol place, just equal to have determined the generator of this modulation symbol.

When 16QAM and 64QAM employing gray mappings, receiving terminal is as long as detect each modulation symbol region, just can determine the generator of each modulation symbol correspondence, and will be specific to each point unlike traditional detection, each point here refers to each constellation point in the modulation constellation.Adopt the PDA of bit-level to detect afterwards, determine the particular location of each modulation symbol again, the flow process of whole algorithm is the progressive clear and definite process of the information of a modulation symbol.The present invention is a kind of thought of progressive detection, in the high-order constellation, is example with 16QAM, and traditional detection is directly to consider 16 constellation point, gets one from 16.The present invention gets 4 (zones of a public generator) earlier in the middle of 16, determine 1 output by the method for PDA again from 4.16 select 4 to compare 16 and select 1 to have certain fault-tolerance.

For 8QAM and this class of 32QAM is not the modulation system of rectangle constellation, their constellation point is respectively the subclass of 16QAM and 64QAM, when detecting, can be used as 16QAM to them earlier and 64QAM detects, determine that concrete constellation positions uses the mode demodulation of 8QAM and 16QAM later on again with the present invention.But these two kinds of modulation systems all are of little use, and also do not do discussing at this.

Secondly, carrying out bit-level PDA detects

The present invention can construct a W to the mimo system under each rank QAM, the various mapping mode, makes x=Wb, and x is the symbolic vector that sends, and b is the bit stream before the modulation.Being taken in the formula (1) has

r＝Hx+n＝HWb+n＝Zb+n (10)

Z＝HW

By formula (10) as can be seen, introduce the afterwards original QAM of W and detect the BPSK detection problem that is converted into an equivalence.

Formula (10) is rewritten as

$r=z_l{b}_l+\underset{k\&NotEqual;l}{\mathrm{\&Sigma;}}{z}_k{b}_k+n$

z _lBe the l row of matrix z, b _kThe k that is former bit stream is individual.

$w=\underset{k\&NotEqual;l}{\mathrm{\&Sigma;}}{z}_k{b}_k+n$

${\mathrm{\&mu;}}_w=E\{\underset{k\&NotEqual;l}{\mathrm{\&Sigma;}}{z}_k{b}_k+n\}=\underset{k\&NotEqual;l}{\mathrm{\&Sigma;}}E\left\{b_k\right\}{z}_k$

$R_w=E\left\{(w-{\mathrm{\&mu;}}_w){(w-{\mathrm{\&mu;}}_w)}^H\right\}=\underset{k\&NotEqual;l}{\mathrm{\&Sigma;}}\mathrm{Var}\left\{b_k\right\}{z}_k{z_k}^H+{\mathrm{\&sigma;}}_n^{2}I$

$K_w=E\left\{(w-{\mathrm{\&mu;}}_w){(w-{\mathrm{\&mu;}}_w)}^T\right\}=\underset{k\&NotEqual;l}{\mathrm{\&Sigma;}}\mathrm{Var}\left\{b_k\right\}{z}_k{z_k}^T$

B wherein _lPossible value is a _j∈ Q={+1 ,-1}, j=1,2

Calculate b according to following flow process _l=a _jProbability

v＝r-a _jz _l-μ _w (11)

$\mathrm{\&phi;}\left(a_j\right)=\exp \left(\begin{array}{c}{-\left\{\begin{array}{c}R\left\{v\right\}\\ I\left\{v\right\}\end{array}\right\}{\mathrm{\&Lambda;}}_j\left\{\begin{array}{c}R\left\{v\right\}\\ I\left\{v\right\}\end{array}\right\}}^T\end{array}\right)---\left(12\right)$

${\mathrm{\&Lambda;}}_j={\left(\begin{array}{cc}R\{R_w+K_w\}& -I\{R_w-K_w\}\\ I\{R_w+K_w\}& R\{R_w-K_w\}\end{array}\right)}^{-1}---\left(13\right)$

$P{r}_1(b_l=a_j|\tilde{r})=\frac{\mathrm{\&phi;}\left(a_j\right)}{\underset{a_j\&Element;Q}{\mathrm{\&Sigma;}}\mathrm{\&phi;}\left(a_j\right)}---\left(14\right)$

The flow process of PDA algorithm is:

The first step is made as i=1 with the iterations counter;

The initialization value is 0.5,

In second step, bit sequence is changed to 1, l=1;

The 3rd step is according to the probable value { P (k) } of current acquisition _{K ≠ l}, utilize formula (11)-(14) to calculate

J=1,2.Institute's value is upgraded corresponding P (l);

If the 4th step is l＜N _TLog ₂M is l=l+1 then, and carries out for the 3rd step, if do not satisfy l＜N _TLog ₂M forwarded for the 5th step to;

The 5th step, if , P (l) has restrained or program has reached maximum iteration time and then jumped to for the 6th step, does not restrain and just continues iteration up to satisfying maximum iteration time;

In the 6th step, this moment, algorithm was restrained, and according to decision rule as follows, each bit was made judgement successively.Adjudicating this bit is 1 or 0.What whole iterative process was calculated is that each bit is got 1 or 0 probability, after iteration finishes, each bit is compared the probability of its value 1 or 0, exports the value of bigger probability correspondence.

l＝1，2，.....N _Tlog ₂M。

N is arranged _TIndividual transmitting antenna, each transmitting antenna sends the symbol that order of modulation is M, so the bit number that sends is N altogether _TLog ₂M.By formula (10) as can be seen, algorithm is intended to set up the relation that receives between vector and the source bits, and traditional detection is to obtain modulation symbol by receiving vector, and modulation symbol is mapped as the source bit again.Directly detect source bits by received signal in the present invention.Can detect N so whenever receive vector by one _TLog ₂M bit.

Among the present invention, bit flows to the process that the modulation symbol vector can be integrated into a linearity.The PDA that goes for the bit-level of the various mapping modes of each rank QAM (quadrature amplitude modulation) detects.

The present invention is under the help of the generator matrix of introducing, and with traditional PDA algorithm at modulation symbol, abbreviation has become the PDA algorithm based on bit stream, thereby has simplified the complexity of computing, and performance also increases.

It is core that the PDA of bit-level detects, and constructs g at different modulation systems and finishes the approach that bit-level PDA detects.For 4QAM, be equivalent to a real part variable, for 16QAM, be equivalent to two real part variablees, for 64QAM, be equivalent to three real part variablees.Linear or nonlinear constellation arrangement mode to other can construct corresponding g, thereby the PDA that finishes bit-level detects.

The value of g is different under different constellation point arrangement modes.With regard to gray mappings, a lot of constellation point arrangement modes is just arranged, but their adjacent constellation point corresponding bit all has only a difference, so all be all kinds of mappings of Gray.Also there is Linear Mapping in this situation.Commonly used arriving in the middle of the communication of the arrangement mode of example among the present invention.

Compare CPDA MIMO detector, the operand of operand of the present invention and CPDA algorithm can be expressed as: the operand that posterior probability of the every calculating of probability number * that the each iteration of C_sum=iterations * will be calculated is paid.

Can observe the iterations approximately equal of two kinds of methods by a large amount of experiments; The posterior probability number of the required calculating of the each iteration of CPDAMIMO detector is MN _T, and the present invention is 2N _TLog ₂M, the posterior probability number that the each iteration of the present invention will be calculated is less; And from formula (11)～(14) as can be seen the present invention calculate a bit the value probability compared to existing technology CPDA calculate the value probability of a complex modulation symbol, the multiplication that contains plural number is less, this has determined operand of the present invention low (the present invention is directed to+1 or-1, CPDA is at complex modulation symbol).So, compare CPDA, operand of the present invention is less, has greatly reduced complexity, especially in the situation of using high-order QAM.Because the BPSK after the conversion detects problem, order of modulation is low, and it is many to send out number of antennas, is more suitable for using PDA to detect, so the present invention also improves to some extent than the CPDA detector performance.

Embodiment

Received signal is from 2 * 2 rayleigh fading channel.According to known order of modulation, mapping mode structure W, utilize W that problem is changed into detection under the BPSK modulation at receiving terminal.Utilize PDA to pursue bit-detection, hereinafter represent with B-PDA.Set different maximum iteration time, use I _MaxExpression.Adopt different order of modulation, mapping mode, experimental result such as Fig. 4, Fig. 5 and shown in Figure 6.The comparison of each detector performance under the 64QAM, the performance of B-PDA slightly is better than CPDA as can be seen.

As can be seen from Figure 4, the present invention is under each order of modulation, and the complexity that each iteration is paid is lower than traditional CPDA, and the complexity that symbolic vector of every detection is paid also is lower than traditional CPDA.

As can be seen from Figure 5, under the 16QAM modulation, performance of the present invention is not worse than traditional CPDA mode and MMSE-OSIC receiver.

As can be seen from Figure 6, under the 64QAM modulation, performance of the present invention is not worse than traditional CPDA mode and MMSE-OSIC receiver.

The ML gray mappings: be to adopt the maximum likelihood probability detection method under the gray mappings, MMSE-OSIC also is a kind of detection mode (based on a least mean-square error).

Description of the invention provides for example with for the purpose of describing, and is not exhaustively or limit the invention to disclosed form.Many modifications and variations are obvious for the ordinary skill in the art.Selecting and describing embodiment is for better explanation principle of the present invention and practical application, thereby and makes those of ordinary skill in the art can understand the various embodiment that have various modifications that the present invention's design is suitable for special-purpose.

Claims (11)

1. at the bit-level probability data aided detection method of high-order QAM multiaerial system, may further comprise the steps:

The bit stream of transmitting terminal is modulated into modulation symbol;

Modulation symbol is diverted on each transmitting antenna, sends to receiving terminal;

Receive modulation symbol by receiving terminal, parse bit stream according to modulation symbol, and carry out the probability data auxiliary detection based on bit stream.

2. according to the described bit-level probability data aided detection method of claim 1 at the high-order QAM multiaerial system, wherein, parse the operation of bit stream according to modulation symbol, comprising:

For shining upon naturally of mapping, the 64QAM modulation system naturally of 4QAM modulation system, 16QAM modulation system, it between modulation symbol and the bit stream linear relationship, modulation symbol is that bit stream and generator matrix multiply each other, and calculates bit stream according to modulation symbol and generator matrix.

3. according to the described bit-level probability data aided detection method of claim 1 at the high-order QAM multiaerial system, wherein, parse the operation of bit stream according to modulation symbol, comprising:

For the gray mappings of 16QAM modulation system, the gray mappings of 64QAM modulation system, between modulation symbol and the bit stream non-linear relation, determine corresponding generator according to modulation symbol, determine corresponding bit stream according to generator again.

4. according to the described bit-level probability data aided detection method of claim 2, wherein at the high-order QAM multiaerial system:

Described generator matrix is relevant with order of modulation, modulation system and number of transmit antennas.

5. according to claim 2 or 4 described bit-level probability data aided detection methods, wherein at the high-order QAM multiaerial system:

For 4QAM modulation system, S=W _4QAMB,

$W_{4\mathrm{QAM}}={\left(\begin{array}{ccccc}{g}_{4\mathrm{QAM}}& .& .& .& 0\\ .& .& & & .\\ .& & .& & .\\ .& & & .& .\\ 0& .& .& .& g_{4\mathrm{QAM}}\end{array}\right)}_{N_T\&times;2N_T};$

For the 16QAM modulation system, under shining upon naturally, S=W _16QAMB,

$W_{16\mathrm{QAM}}={\left(\begin{array}{ccccc}{g}_{16\mathrm{QAM}}& .& .& .& 0\\ .& .& & & .\\ .& & .& & .\\ .& & & .& .\\ 0& .& .& .& g_{16\mathrm{QAM}}\end{array}\right)}_{N_T\&times;4N_T};$

For the 64QAM modulation system, under shining upon naturally, S=W _64QAMB,

$W_{64\mathrm{QAM}}={\left(\begin{array}{ccccc}{g}_{64\mathrm{QAM}}& .& .& .& 0\\ .& .& & & .\\ .& & .& & .\\ .& & & .& .\\ 0& .& .& .& g_{64\mathrm{QAM}}\end{array}\right)}_{N_T\&times;6N_T}$

Wherein, S represents modulation symbol, and B represents bit stream, and W represents generator matrix, N _TIt is the number of transmit antennas of system.

6. according to the described bit-level probability data aided detection method of claim 3, wherein at the high-order QAM multiaerial system:

For the 16QAM modulation system, under gray mappings, determine corresponding generator according to the quadrant at modulation symbol place;

For the 64QAM modulation system, under gray mappings, determine corresponding generator according to the zone on the planisphere plane at modulation symbol place.

7. at the bit-level probability data auxiliary detection device of high-order QAM multiaerial system, comprising:

Transmitting terminal is modulated into modulation symbol with bit stream; Modulation symbol is diverted on each transmitting antenna, sends to receiving terminal;

Receiving terminal receives modulation symbol, parses bit stream according to modulation symbol, and carries out the probability data auxiliary detection based on bit stream.

8. according to the described bit-level probability data auxiliary detection device of claim 7, wherein at the high-order QAM multiaerial system:

For shining upon naturally of mapping, the 64QAM modulation system naturally of 4QAM modulation system, 16QAM modulation system, it between modulation symbol and the bit stream linear relationship, modulation symbol is that bit stream and generator matrix multiply each other, and receiving terminal calculates bit stream according to modulation symbol and generator matrix;

For the gray mappings of 16QAM modulation system, the gray mappings of 64QAM modulation system, between modulation symbol and the bit stream non-linear relation, receiving terminal is determined corresponding generator according to modulation symbol, determines corresponding bit stream according to generator again.

9. according to claim 7 or 8 described bit-level probability data auxiliary detection devices, wherein at the high-order QAM multiaerial system:

For 4QAM modulation system, S=W _4QAMB,

$W_{4\mathrm{QAM}}={\left(\begin{array}{ccccc}{g}_{4\mathrm{QAM}}& .& .& .& 0\\ .& .& & & .\\ .& & .& & .\\ .& & & .& .\\ 0& .& .& .& g_{4\mathrm{QAM}}\end{array}\right)}_{N_T\&times;2N_T};$

For the 16QAM modulation system, under shining upon naturally, S=W _16QAMB,

$W_{16\mathrm{QAM}}={\left(\begin{array}{ccccc}{g}_{16\mathrm{QAM}}& .& .& .& 0\\ .& .& & & .\\ .& & .& & .\\ .& & & .& .\\ 0& .& .& .& g_{16\mathrm{QAM}}\end{array}\right)}_{N_T\&times;4N_T};$

For the 64QAM modulation system, under shining upon naturally, S=W _64QAMB,

$W_{64\mathrm{QAM}}={\left(\begin{array}{ccccc}{g}_{64\mathrm{QAM}}& .& .& .& 0\\ .& .& & & .\\ .& & .& & .\\ .& & & .& .\\ 0& .& .& .& g_{64\mathrm{QAM}}\end{array}\right)}_{N_T\&times;6N_T};$

Wherein, S represents modulation symbol, and B represents bit stream, and W represents generator matrix, N _TIt is the number of transmit antennas of system;

For the 16QAM modulation system, under gray mappings, determine corresponding generator according to the quadrant at modulation symbol place;

For the 64QAM modulation system, under gray mappings, determine corresponding generator according to the zone on the planisphere plane at modulation symbol place.

10. at the receiving terminal of the bit-level probability data auxiliary detection of high-order QAM multiaerial system, comprising:

Receiver module is used to receive modulation symbol;

Parsing module is coupled in receiver module, is used for parsing bit stream according to modulation symbol;

Detection module is coupled in parsing module, is used for carrying out the probability data auxiliary detection based on bit stream.

11. according to the receiving terminal of the described bit-level probability data auxiliary detection at the high-order QAM multiaerial system of claim 10, wherein:

For shining upon naturally of mapping, the 64QAM modulation system naturally of 4QAM modulation system, 16QAM modulation system, it between modulation symbol and the bit stream linear relationship, modulation symbol is that bit stream and generator matrix multiply each other, and parsing module calculates bit stream according to modulation symbol and generator matrix;

For the gray mappings of 16QAM modulation system, the gray mappings of 64QAM modulation system, between modulation symbol and the bit stream non-linear relation, parsing module is determined corresponding generator according to modulation symbol, determines corresponding bit stream according to generator again.

Images