CN101854328B - Layered space-time nonlinear precoding method in multi-carrier code division multiple access (MC-CDMA) system - Google Patents

Abstract

The invention relates to a layered space-time nonlinear precoding method in a multi-carrier code division multiple access system. In the method, a precoding system is firstly established, and the feedback channel information in the system is decomposed into a product of a unitary matrix and an upper triangular matrix by using a geometric mean value. Obtain the precoding matrix with the same equivalent noise gain, and then perform nonlinear modular algebraic precoding THP between the MC-CDMA subcarrier channels at the transmitting end, and then compare the subcarrier signals after THP with the channel geometric mean decomposition The unitary matrix is multiplied, and then sent out from the corresponding antenna; at the receiving end, the zero-forcing ZF criterion or the minimum mean square error MMSE criterion is used for processing; this method effectively eliminates the error-layer transmission effect of the layered space-time code, and improves the system The bit error performance reduces the complexity of the downlink receiver, and can effectively resist channel frequency selective fading, thereby improving the transmission performance of the system.

Description

Layered space-time nonlinear method for precoding in the MC-CDMA system

Technical field

The present invention relates to layered space-time nonlinear method for precoding in the MC-CDMA system, belong to the multiplexing layering of multiple input multiple output multicarrier system space lift-off technology field when empty.

Background technology

The multiple-input and multiple-output MIMO technology of broad sense relates to extensively, mainly comprises transmit diversity techniques and space multiplexing technique.Wherein space multiplexing technique is the different information of emission on different antennas, and the V-BLAST sign indicating number of AT&T Labs is the typical application of space multiplexing technique; And transmit diversity techniques is that emission comprises the signal of same information on different antennas, thereby reaches the effect of space diversity.Technological based on the Space Time Coding of transmit diversity because transmit diversity on the spatial domain and the coding on the time-domain are combined, can overcome channel fading largely, improve systematic function, thereby receive much concern.

At present, main three kinds of multicarrier CDMA scheme, MC/DS-CDMA, MT-CDMA and MC-CDMA.MC/DS-CDMA is proposed by V.Dasilva of University of Toronto and E.S.Sousal; It is that the transmission data sequence is at first become multidiameter delay output through serial to parallel conversion; Parallel then every circuit-switched data is carried out the OFDM modulation to this multichannel data at last by identical short spreading code spread spectrum; MT-CDMA is proposed by Belgian L.Vandendorpe; It also is to utilize given frequency expansion sequence in time domain, to expand the data flow behind serial to parallel conversion; Different with the former is that MT-CDMA adopts and the proportional long frequency expansion sequence of carrier frequency, makes system can hold more user; And MC-CDMA to be Fazel, Papke by the Linnartz of the U.S., Yee, Fettweis and Germany the earliest independently propose separately; It is that each information symbol carries out spread spectrum by a specific spread-spectrum code chip; Then each spread-spectrum signal is modulated on the corresponding subcarrier; Therefore, if the length of spreading code is N, what this so corresponding N number of sub-carrier was transmitted is identical information data.Wherein MC-CDMA has best spectrum distribution, and antijamming capability is strong, and the realization of transmitter is simpler.

The multiple-input and multiple-output mimo system adopts the Space Time Coding technology can greatly improve the capacity of wireless communication system, and effectively improves the performance of wireless communication system.MC-CDMA MC-CDMA scheme combines the advantage of CDMA and OFDM (modulation of Orthogonal Frequency Division Multiplexing OFDM) technology, can resist the intersymbol interference that is caused by the frequency selectivity multipath channel effectively.MIMO is combined with the MC-CDMA technology, can further improve the performance and the transmission rate of system.For further improving the system receiver performance and reducing its complexity; Novel signal processing technology becomes the focus of research in the industry; Precoding technique is exactly to utilize complete channel state information or partial channel-state information to carry out corresponding preliminary treatment to sending symbol at transmitting terminal, is the modern signal processing technology of purpose to improve power system capacity or reduction error rate of system etc.

At present; The research that in the MIMO-MC-CDMA system, detects to the vertical layered space-time code VBLAST that is proposed by the AT&T Labs mainly concentrates on further raising detector performance and reduces on the complexity, and traditional detection method mainly contains zero method (ZF), the least mean-square error method (MMSE) etc. of compeling.The MIMO-MC-CDMA system is based on ordering iteration interference elimination method, to obtain more performance; Based on low-complexity MIMO-MC-CDMA receiver, obtain corresponding performance improvement simultaneously.Though above-mentioned two kinds of detection methods have obtained should improve mutually at receiver performance, shortcoming is that complexity height and performance are not optimum, has mistake layer transmission effects, and the complexity of receiver improves along with performance improvement simultaneously.Can overcome the shortcoming of traditional receiving terminal detection method based on the transmitting end channel precoding technique, adopt linear predictive coding, though can reduce the complexity of receiver, but still can not eliminate a mistake layer transmission effects; Adopt non-linear THP precoding; Not only reduced the complexity of receiver; And effectively eliminate mistake layer transmission effects, but traditional THP method for precoding can be the triangle channel with different diagonal element values with channel decomposing mostly based on QR triangle decomposition method; Make subchannel have different gains, thereby cause certain performance loss; Decompose pre-coding scheme based on geometric mean,, make gains such as subchannel acquisition because all diagonal element values are equal to the geometric mean of channel matrix characteristic value; The present invention is incorporated into the MIMO-MC-CDMA layering when empty in the scheme with the thought of nonlinear precoding; And the non-linear modular algebra THP of VBLAST method for precoding in a kind of MIMO-MC-CDMA downlink system proposed; Simulation result shows that method proposed by the invention effectively reduces the error performance of system than conventional method, and reduces the downlink receiver complexity.

Summary of the invention

The purpose of this invention is to provide layered space-time nonlinear method for precoding in a kind of MC-CDMA system, the scheme traditional detection exists mistake layer transmission effects and the high problem of decoding complexity when solving in the MIMO-MC-CDMA system vertical layered space.

For realizing above-mentioned purpose; Layered space-time nonlinear method for precoding technical scheme is following in the MC-CDMA system of the present invention; 1. this method is at first set up pre-coding system; Adopt geometric mean to be decomposed into the product of unitary matrice and upper triangular matrix to feedback channel information in this system then, the pre-coding matrix that acquisition has the same equivalent noise gain carries out non-linear modular algebra precoding THP again between transmitting terminal MC-CDMA MC-CDMA sub-carrier channels; Again each is multiplied each other through sub-carrier signal behind the non-linear modular algebra precoding THP and the unitary matrice that is obtained by the decomposition of channel geometric mean, go out from corresponding antenna transmission then; Adopt urgent zero ZF criterion or least mean-square error MMSE criterion to handle at receiving terminal.

[0008]Further, said pre-coding system, establishing total number of users is U, number of transmit antennas is N, the reception antenna number be M (

); At transmitting terminal; Each user data is after relevant treatment such as vertical layered space-time code VBLAST mapping, coding; Form N layer parallel data, more every layer data is carried out spread spectrum and form data, then each layer data is carried out serial to parallel conversion, IFFT conversion, adds Cyclic Prefix, carried out precoding, and becomes string and be modulated on each transmitting antenna; At receiving terminal; Inverse operation such as each receives the signal serial to parallel conversion, remove Cyclic Prefix, FFT; The frequency selectivity multidiameter fading channel just is converted into the subchannel of Nc flat fading, and then the signal of all u users of M root reception antenna after despreading on the k subcarrier can be expressed as:

; Wherein, Equality the right first is expectation user profile; Second portion is the distracter of other user to its generation, and third part is a white Gaussian noise;

;

and

is respectively transmission symbolic vector, receiving symbol vector and white Gaussian noise on the k number of sub-carrier;

is N on the complex field

N pre-coding matrix;

is M on the complex field

N power adjustment receiving matrix;

is M on the complex field

N matrix; Its element

(n=1 ... N; M=1 ... M) on the expression k number of sub-carrier from transmitting antenna n to the channel frequency response coefficient reception antenna m,

is u the spread-spectrum code chip on the user k number of sub-carrier.

Said geometric mean decomposable process is:

geometric mean of matrix H of dimension decompose GMD and can be decomposed into

; Q, P are unitary matrice; The real main track upper triangular matrix of dimension that R is

; K=rank (X); Element on the diagonal of the equal matrix R of its leading diagonal element equals the geometric mean of all K the positive singular values of X when being equivalent; Satisfy

, .

Said modular algebra precoding THP method at first is that transmitting terminal carries out constellation mapping to the input data, asks modular arithmetic through feedback then, and channel adds noise again, and receiving terminal carries out handled mould operation judgement again.

The said mould of asking is operated mathematic(al) representation for

wherein;

is for sending signal; K=1; 2; ..., N.

is modulus; Real part is got in Re () expression; Imaginary part is got in Im () expression;

expression rounds downwards; The purpose of asking mould is mainly the power that sends signal x to be retrained; After promptly asking modular arithmetic; Satisfy

; Signal no longer is original constellation point; Get in the zone but be evenly distributed on the square of the length of side, and have the same distribution space with former constellation for

.

Said non-linear modular algebra GMD-THP-ZF method for precoding testing process is:

(1) channel decomposing: at first; Suppose that number of transmit antennas equals the reception antenna number; H is matrix; Channel matrix H is decomposed then through GMD:

; Wherein, Q, P are unitary matrice; The R upper triangular matrix; Feedforward matrix

; Weighting matrix

, feedback matrix B=GR;

(2) precoding feedback and ask modular arithmetic:

Wherein, X=c-(B-I) x+d; C+d=Bx;

, v=c+d;

(3) receiving terminal receives signal and processing:

， ，

， ，

，

， ，

；

(4) carry out modular arithmetic and judgement again:

， ，

Wherein, Channel

; Noise n;

; Wherein, Q is a unitary matrice; The R upper triangular matrix; Feedforward matrix

, weighting matrix , feedback matrix B=GR.

The basic thought of said modular algebra precoding based on MMSE-THP is; Satisfy in the transmitting terminal transmitted power under the condition of constraint; The data message error that data message that receiving terminal receives and transmitting terminal are sent is minimum;

; Can get:

, is signal power to noise power ratio.

After said receiving terminal receives signal, receiving signal through receiving the adaptive power preliminary treatment, decoding and despreading are handled when carrying out the layering sky again, adjudicate to received signal at last, recover to transmit.

Layered space-time nonlinear method for precoding in the MC-CDMA system of the present invention; Can eliminate the mistake layer transmission effects of hierarchical space-time code effectively; Improved system's error performance; Reduce the complexity of descending link receiver, can resist the selectivity of channel frequency decline effectively, thereby improved the transmission performance of system.

Description of drawings

Fig. 1 is a VBLAST-MC-CDMA pre-coding system model of the present invention;

Fig. 2 is a THP method for precoding structured flowchart;

Fig. 3 is that 22 of multi-users receive ZF criterion error rate analogous diagram;

Fig. 4 is that 44 of multi-users receive ZF criterion error rate analogous diagram;

Fig. 5 is that 22 of multi-users receive MMSE criterion error rate analogous diagram;

Fig. 6 is that 44 of multi-users receive MMSE criterion error rate analogous diagram.

Embodiment

1, system model

VBLAST-MC-CDMA pre-coding system model is as shown in Figure 1; Suppose that total number of users is U; Number of transmit antennas is N, and the reception antenna number is M ( ).At transmitting terminal; Each user data is after relevant treatment such as VBLAST mapping, coding; Form N layer parallel data; Again every layer data is carried out spread spectrum and form data, then each layer data is carried out serial to parallel conversion, IFFT conversion, adds Cyclic Prefix, carried out precoding, and becomes string and be modulated on each transmitting antenna.At receiving terminal, inverse operation such as each receives the signal serial to parallel conversion, removes Cyclic Prefix, FFT.The frequency selectivity multidiameter fading channel just is converted into the subchannel of Nc flat fading like this, and then the signal of all u users of M root reception antenna after despreading on the k subcarrier can be expressed as:

(1)

Wherein, equality the right first is expectation user profile, and second portion is the distracter of other user to its generation, and third part is a white Gaussian noise;

;

and

is respectively transmission symbolic vector, receiving symbol vector and white Gaussian noise on the k number of sub-carrier;

is N on the complex field

N pre-coding matrix;

is M on the complex field N power adjustment receiving matrix;

is M on the complex field

N matrix; Its element

(n=1 ... N; M=1 ... M) on the expression k number of sub-carrier from transmitting antenna n to the channel frequency response coefficient reception antenna m, is u the spread-spectrum code chip on the user k number of sub-carrier.

2, nonlinear precoding method:

2.1 modular algebra THP

THP (Tomlinson-Harashima Precoding) precoding is a kind of typical nonlinear computing, and its basic thought can improve systematic function effectively through carrying out non-linear preliminary treatment at transmitting terminal according to the channel information state.THP method for precoding structured flowchart is as shown in Figure 2.At first transmitting terminal carries out constellation mapping to the input data, asks modular arithmetic through feedback then, and channel adds noise again, and receiving terminal carries out handled mould operation judgement again.The transmitting terminal of modular algebra pre-coding system by mapping, ask modular arithmetic, feedback matrix B to form, receiving terminal is by feedforward matrix F, weighting matrix G, and corresponding with the transmitting terminal modular arithmetic composition of asking.

The modular algebra theory is the typical nonlinear method, asks mould operation mathematic(al) representation:

; Wherein,

is for sending signal; K=1; 2 ..., N.

is modulus; Real part is got in Re () expression; Imaginary part is got in Im () expression;

expression rounds downwards; The purpose of asking mould is mainly the power that sends signal x to be retrained; After promptly asking modular arithmetic; Satisfy

; Signal no longer is original constellation point; Get in the zone but be evenly distributed on the square of the length of side, and have the same distribution space with former constellation for

.

2.2 QR-THP method for precoding

Suppose that transmitting terminal can accurately know channel condition information, non-linear modular algebra QR-THP-ZF method for precoding testing process is:

(1) channel decomposing: at first; Be simple and Convenient Calculation; Suppose that number of transmit antennas equals the reception antenna number; H is

matrix; Channel matrix H is decomposed then through QR:

; Wherein, Q is a unitary matrice; The R upper triangular matrix; Feedforward matrix

, weighting matrix , feedback matrix B=GR;

(2) precoding feedback and ask modular arithmetic:

Wherein, X=c-(B-I) x+d; C+d=Bx;

, v=c+d;

(3) receiving terminal receives signal and processing:

，

，

，

，

，

，

，

；

(4) carry out modular arithmetic and judgement:

，

，

Wherein, H=QR; Q is a unitary matrice; The R upper triangular matrix; Feedforward matrix ; Weighting matrix

; Wherein

, feedback matrix B=GR=GFH.Basic thought based on the modular algebra precoding of MMSE-THP is, satisfies in the transmitting terminal transmitted power under the condition of constraint, the data message error minimum that data message that receiving terminal receives and transmitting terminal are sent. ; Can get:

,

is signal power to noise power ratio.Was zero while based on ZF criterion method for precoding putting other signals interference by force; Also lost a part of utilizable information; Whether relatively responsive the ZF criterion accurately to the channel information state, reduced systematic function to a certain extent, and equilibrium has been considered characteristic of channel matrix and received noise in the signal to the influence of systematic function based on MMSE criterion method for precoding; Alleviate the influence of mistake layer transmission effects to a certain extent, thereby improved systematic function.

2.3 GMD-THP method for precoding

The non-linear modular algebra THP method for precoding of tradition is based on the QR triangle decomposition mostly; Can be triangle channel with channel decomposing with different diagonal element values; Make subchannel have different gains, thereby cause certain performance loss, decompose pre-coding scheme based on geometric mean; All diagonal element values are equal to the geometric mean of channel matrix characteristic value, make gains such as subchannel acquisition.

According to matrix theory knowledge; The geometric mean decomposition (GMD) of supposing the matrix H of any matrix

dimension can be decomposed into

; Q; P is unitary matrice; The real main track upper triangular matrix of dimension that R is

; K=rank (X); Element on the diagonal of the equal matrix R of its leading diagonal element equals the geometric mean of all K the positive singular values of X when being equivalent; Satisfy

, .On the basis of singular value decomposition, can derive geometric average through displacement transformation and bilateral Givens conversion and decompose.

Suppose that transmitting terminal can accurately know channel condition information, non-linear modular algebra GMD-THP-ZF method for precoding testing process is:

(1) channel decomposing: at first; Be simple and Convenient Calculation; Suppose that number of transmit antennas equals the reception antenna number; H is

matrix; Channel matrix H is decomposed then through GMD:

; Wherein, Q; P is a unitary matrice, R upper triangular matrix, feedforward matrix ; Weighting matrix

, feedback matrix B=GR;

(2) precoding feedback and ask modular arithmetic:

Wherein, X=c-(B-I) x+d; C+d=Bx; , v=c+d;

(3) receiving terminal receives signal and processing:

，

，

，

，

，

，

，

；

(4) carry out modular arithmetic and judgement again:

，

，

Wherein, Channel

; Noise n;

; Wherein, Q is a unitary matrice; The R upper triangular matrix; Feedforward matrix

, weighting matrix

, feedback matrix B=GR.Basic thought based on the modular algebra precoding of MMSE-THP is, satisfies in the transmitting terminal transmitted power under the condition of constraint, the data message error minimum that data message that receiving terminal receives and transmitting terminal are sent. ; Can get:

,

is signal power to noise power ratio.

Emulation relatively

The system emulation parameter is provided with as follows: multi-user's number (4); Multiple access disturbs (MAI=10dB); Antenna number (receive for 44, receive for 6 6); Sub-carrier number (64); Cyclic Prefix (20%); Modulation system (QPSK); Channel circumstance (frequency selectivity multipath Rayleigh channel, multipath number: 4 footpaths, symbol of every footpath time-delay).

Provided respectively among Fig. 3 and Fig. 4 in 44 receipts and 66 error rates of receiving under the employing ZF criterion situation, the present invention carries out emulation to the urgent zero method of VBLAST system linear, urgent zero method, linear ZF method for precoding, non-linear modular algebra QR-THP-ZF method for precoding and GMD-THP-ZF method for precoding proposed by the invention and compares in the MIMO-MC-CDMA system.Abscissa among the figure is represented signal to noise ratio, and unit is dB, and ordinate is represented bit error rate.Can know that from simulation result linear ZF method and linear ZF method for precoding have close performance of BER, but the receiver of linear ZF method for precoding has the low complexity that receives; The performance of BER that non-linear GMD-THP compels zero method is much better than traditional Z F detection method and ordering ZF detection method; Have than non-linear QR-THP and compel the zero more excellent error performance of method; Having reduced the error rate to a certain extent, is the performance improvement that 3-4dB is arranged on order of magnitude in the error rate.

Provided respectively among Fig. 5 and Fig. 644 receipts and received the error rates that adopt under the MMSE criterion situation with 66, the non-linear GMD-THP-MMSE method for precoding to the Nonlinear M MSE of VBLAST system method, linear MMSE method, ordering MMSE method, linear MMSE method for precoding and non-linear QR-THP-MMSE method for precoding and the present invention's proposition carries out emulation relatively in the MIMO-MC-CDMA system respectively.Abscissa among the figure is represented signal to noise ratio, and unit is dB, and ordinate is represented bit error rate.Can know that from simulation result linear MMSE method and linear MMSE method for precoding have close performance of BER, but the receiver of linear MMSE method for precoding has the low complexity that receives; The performance of BER of non-linear GMD-THP-MMSE method is much better than traditional MMSE detection method and ordering MMSE detection method; Have than the more excellent error performance of non-linear QR-THP-MMSE method; Having reduced the error rate to a certain extent, is the performance improvement that 2-3dB is arranged on

order of magnitude in the error rate.

The practical implementation step:

One, transmitting terminal

1, supposes 22 of multi-users are received MIMO multicarrier system, at first user u input information is carried out the layering mapping:

;

2, know by Fig. 1; For u user; Every transmitting antenna carries out spread spectrum: ; Wherein,

is u the spread-spectrum code chip on the user k number of sub-carrier, and Nc is the length of spreading code;

3, decomposition is decomposed into feedback channel according to geometric mean:

, wherein, Q, PBe unitary matrice, RUpper triangular matrix, feedforward matrix

, weighting matrix

, feedback matrix B=GR

4, non-linear modular algebra precoding can be known by Fig. 2, and the expression formula of modular algebra feedback computing is:

，

，

，

；

5, hierarchical signal behind the spread spectrum and feedback channel pre-coding matrix

are multiplied each other, then signal phasor is expressed as on u pairing k number of sub-carrier of user:

;

6, every layer signal carries out the OFDM modulation, launches from corresponding antenna then to get final product.

Two, receiving terminal

1, can know by formula (1), receive signal and form by desired user signal, interference user signal and white Gaussian noise.The reception signal phasor on u user k number of sub-carrier then:

；

2, receiving signal through receiving the adaptive power preliminary treatment, following formula becomes:

3, decoding when carrying out the layering sky again, following formula becomes:

4, vector carries out corresponding non-linear modular algebra processing to received signal: ;

5, dock collection of letters signal at last and carry out despreading and judgement, recover to transmit:

Wherein Dec (.) expression judgement computing.

Claims (8)

1. layered space-time nonlinear method for precoding in the MC-CDMA system; It is characterized in that: this method is at first set up pre-coding system; Adopt geometric mean to be decomposed into the product of unitary matrice and upper triangular matrix to feedback channel information in this system then; Acquisition has the pre-coding matrix of same equivalent noise gain; Between transmitting terminal MC-CDMA MC-CDMA sub-carrier channels, carry out non-linear modular algebra precoding THP again, again each is multiplied each other through sub-carrier signal behind the non-linear modular algebra precoding THP and the unitary matrice that is obtained by the decomposition of channel geometric mean, go out from corresponding antenna transmission then; Adopt urgent zero ZF criterion or least mean-square error MMSE criterion to handle at receiving terminal.

2. nonlinear precoding method according to claim 1 is characterized in that: said pre-coding system, and establishing total number of users is U; Number of transmit antennas is N; The reception antenna number is M, and N≤M, at transmitting terminal; Each user data is after vertical layered space-time code VBLAST mapping, encoding process; Form N layer parallel data, more every layer data is carried out spread spectrum and form data, then each layer data is carried out serial to parallel conversion, IFFT conversion, adds Cyclic Prefix, carried out precoding, and becomes string and be modulated on each transmitting antenna; At receiving terminal; Each receives the signal serial to parallel conversion, remove Cyclic Prefix, the FFT operation; The frequency selectivity multidiameter fading channel just is converted into the subchannel of Nc flat fading, and then the signal of all u users of M root reception antenna after despreading on the k subcarrier can be expressed as:

$Y^{\mathrm{Uk}}=G^{\mathrm{Uk}}{H}^{\mathrm{Uk}}{F}^{\mathrm{Uk}}{X}^{\mathrm{Uk}}+\underset{i=1,u\≠i}{\overset{U}{\mathrm{\Σ}}}{c}^{\mathrm{Uk}}{c}^{\mathrm{Jk}}{G}^{\mathrm{Uk}}{H}^{\mathrm{Uk}}{F}^{\mathrm{Uk}}{X}^{\mathrm{Uk}}+c^{\mathrm{Uk}}{G}^{\mathrm{Uk}}{W}^{\mathrm{Uk}},$ Wherein, equality the right first is expectation user profile, and second portion is the distracter of other user to its generation, and third part is a white Gaussian noise; $X^{\mathrm{Uk}}={[X_1^{\mathrm{Uk}}\·\·\·X_N^{\mathrm{Uk}}]}^T,$ $Y^{\mathrm{Uk}}={[Y_1^{\mathrm{Uk}}\·\·\·Y_M^{\mathrm{Uk}}]}^T$ With $W^{\mathrm{Uk}}={[W_1^{\mathrm{Uk}}\·\·\·W_M^{\mathrm{Uk}}]}^T$ Be respectively transmission symbolic vector, receiving symbol vector and white Gaussian noise on the k number of sub-carrier,

Be N on the complex field * N pre-coding matrix,

Be M on the complex field * N power adjustment receiving matrix,

Be M on the complex field * N matrix, its element

(n=1 ... N, m=1 ... M) on the expression k number of sub-carrier from transmitting antenna n to the channel frequency response coefficient reception antenna m, c ^UkBe u the spread-spectrum code chip on the user k number of sub-carrier.

3. nonlinear precoding method according to claim 2 is characterized in that: said geometric mean decomposable process is: the geometric mean of the matrix H of M * N dimension decomposes GMD and can be decomposed into H=QRP ^H, Q, P are unitary matrice, and R is the upper triangular matrix of K * K dimension, K=rank (X), the element on the diagonal of matrix R is equivalent, equals the geometric mean of all K the positive singular values of X simultaneously.

4. nonlinear precoding method according to claim 1; It is characterized in that: said non-linear modular algebra precoding THP method at first is that transmitting terminal carries out constellation mapping to the input data; Ask modular arithmetic through feedback then; Channel adds noise again, and receiving terminal carries out handled, carries out modular arithmetic and judgement again.

5. nonlinear precoding method according to claim 4 is characterized in that: the said mould operation mathematic(al) representation of asking does

Wherein, x _kBe the transmission signal, k=1,2 ..., N; τ is a modulus, and real part is got in Re () expression, and imaginary part is got in Im () expression,

Expression rounds downwards, and the purpose of asking mould is mainly the power that sends signal x to be retrained.

6. according to claim 4 or 5 described nonlinear precoding methods, it is characterized in that: said non-linear modular algebra method for precoding testing process is:

(1) channel decomposing: at first, suppose that number of transmit antennas equals the reception antenna number, H is N * N matrix, and channel matrix H is decomposed then through GMD: H=QRP ^H, wherein, Q, P are unitary matrice, R upper triangular matrix, feedforward matrix F=Q ^H, weighting matrix G=diag (1/r ₁₁..., 1/r _NN), feedback matrix B=GR;

(2) precoding feedback and ask modular arithmetic: $x={\mathrm{Mod}}_{2\sqrt{M}}(c-(B-I)x),$

Wherein, x=c-(B-I) x+d, c+d=Bx, x=B ^-1V, v=c+d;

(3) receiving terminal receives signal and processing:

$\hat{c}=Q\left({\mathrm{mod}}_{2\sqrt{M}}\left(r\right)\right),$

$\hat{c}=Q\left({\mathrm{mod}}_{2\sqrt{M}}\left({\mathrm{GQ}}^{H}y\right)\right),$

$\hat{c}=Q\left({\mathrm{mod}}_{2\sqrt{M}}\left({\mathrm{GQ}}^H(\mathrm{Hx}+n)\right)\right),$

$\hat{c}=Q\left({\mathrm{mod}}_{2\sqrt{M}}\left({\mathrm{GQ}}^H({\mathrm{HPB}}^{-1}v+n)\right)\right),$

$\hat{c}=Q\left({\mathrm{mod}}_{2\sqrt{M}}\left({\mathrm{GQ}}^H({\mathrm{HPB}}^{-1}v+n)\right)\right),$

$\hat{c}=Q\left({\mathrm{mod}}_{2\sqrt{M}}\left({\mathrm{GQ}}^H({\mathrm{QRP}}^H{\mathrm{PB}}^{-1}v+n)\right)\right),$

$\hat{c}=Q\left({\mathrm{mod}}_{2\sqrt{M}}(v+n^{\′})\right),$

$\hat{c}=Q\left({\mathrm{mod}}_{2\sqrt{M}}(c+d+n^{\′})\right);$

(4) carry out modular arithmetic and judgement again:

$\hat{c}={\mathrm{mod}}_{2\sqrt{M}}\left(Q\left({\mathrm{mod}}_{2\sqrt{M}}(c+d+n^{\′})\right)\right),$

$\hat{c}=Q(c+n^{\′}),$

Wherein, channel H=QRP ^H, noise n, H=QR, wherein, Q is a unitary matrice, R upper triangular matrix, feedforward matrix F=Q ^H, weighting matrix G=diag (1/r ₁₁..., 1/r _NN), feedback matrix B=GR.

7. nonlinear precoding method according to claim 5; It is characterized in that: said employing least mean-square error MMSE criterion is handled; Be to satisfy under the condition of constraint in the transmitting terminal transmitted power, the data message error minimum that data message that receiving terminal receives and transmitting terminal are sent

Can get: QRP ^H=(HH ^H+ σ ²I) H ^-H, σ ²Be signal power to noise power ratio.

8. nonlinear precoding method according to claim 6; It is characterized in that: after said receiving terminal receives signal, receiving signal through receiving the adaptive power preliminary treatment, decoding and despreading are handled when carrying out the layering sky again; Adjudicate to received signal at last, recover to transmit.

Images