CN103209051B - The two step method for precoding of a kind of coordinate multipoint joint transmission system under multi-user scene - Google Patents

Abstract

The invention discloses the two step method for precoding of a kind of coordinate multipoint joint transmission system under multi-user scene, first by interference registration process, the optimization aim based on minimizing the interference of each user to other each users and leaking is set up at single base station end, and adopt alternating minimization method, obtain the first step pre-coding matrix for each user and receiving matrix respectively; Secondly, the receiving matrix for each user first step obtained, first step pre-coding matrix and useful channel matrix are combined into equivalent channel matrix, and corresponding phase rotation matrix is set on this basis, interference between each user's different antennae is converted into useful signal, effectively utilize each base station to the transmitting power of this user, and obtain second step pre-coding matrix and final pre-coding matrix.The error rate of the present invention declines obvious, accelerate the downward trend of ber curve, and system Mean Speed is also significantly improved.

Description

The two step method for precoding of a kind of coordinate multipoint joint transmission system under multi-user scene

Technical field:

The invention belongs to wireless communication field, be specifically related to the two step method for precoding of a kind of coordinate multipoint joint transmission system under multi-user scene.

Background technology:

Joint transmission technology in Synergistic multi-point system is by the cell cooperative service in cooperative cluster, thus enable multiple base station transmit data to user simultaneously, presence of intercell interference is converted into useful signal, the receptivity of service-user can be significantly improved, and then promote the spectrum efficiency of cell edge.But when there is multiple service-user in cooperative cluster, significant inter-user interference can cause the heavy losses of joint transmission systematic function; In addition, when each user configures many reception antennas and receives multiple data flow simultaneously, between data flow, also interference can be produced.For suppressing system interference, need to adopt corresponding pre-coding scheme.

As the channel condition information (channelstateinformation of all service-users, when CSI) sharing in cooperative cluster, overall method for precoding can be used, the channel matrix of different base station to unique user is integrated, carry out overall precoding, the method has good systematic function; But its computation complexity is higher, and often need the different channels information of a center processing unit to user to process, specific implementation is comparatively difficult.Relatively, in local method for precoding, each base station obtains the channel information between itself and different user, or shared CSI information additional certain between base station, and base station utilizes these information to carry out precoding; This method avoid centralized processing mode, reduce the difficulty of computation complexity and technology enforcement, but, the performance of local method for precoding may by the limited impact (block diagonalization (regularblockdiagonalization of such as traditional specification of available space resources, RBD) precoding and based on letter leakage make an uproar than (signal-to-leakage-plus-noiseratio, SLNR) maximized precoding can be subject to system antenna configuration or space resources utilizes insufficient restriction), thus there is the reduction of precoding performance.

In sum, for coordinate multipoint joint transmission system, design between a kind of rational suppression user and the method for precoding that disturbs between single user different data streams is necessary.

Summary of the invention:

For above problem, the present invention proposes the two step method for precoding of a kind of coordinate multipoint joint transmission system under multi-user scene, can suppress to disturb between user and between single user different data streams, and be converted into useful signal, the bit error rate performance of user can be improved and the transmission rate of elevator system.

The present invention is achieved through the following technical solutions:

The two step method for precoding of coordinate multipoint joint transmission system under multi-user scene, comprise the following steps:

1) by interference registration process, inter-user interference overlap suffered by each user is placed in the orthogonal intersection space receiving space, and use alternating minimization method at the first step pre-coding matrix of each user of base station end iteration optimization and receiving matrix, enable user receive glitch-free useful signal;

2) on the basis that inter-user interference suppresses, each base station obtains the equivalent channel matrix for different user, arranges phase rotation coefficient, and interference between the data flow suffered by each user is converted into useful signal, effectively utilizes base station transmitting power to improve systematic function.

In described interference registration process, the object representation of interference registration process is:

${\mathrm{\Φ}}_k^H{H}_{k,i}{W}_{l,i}^{\left(a\right)}=0,\∀l\≠k,i;$

$\mathrm{rank}\left({\mathrm{\Φ}}_k^H{H}_{k,i}{W}_{k,i}^{\left(a\right)}\right)=L_k,\∀k,i;$

Wherein represent the first step pre-coding matrix of i-th base station to a kth user, rank represents acquisition rank of matrix; Φ _kfor the receiving matrix of user k, H represents conjugate transpose; H _k,ibe the channel matrix of i-th base station to a kth user, L _kfor the transmitting data stream number of a kth user.

Adopt alternating minimization method to obtain the suboptimal solution of the target disturbing registration process, set up following optimization aim:

$\min \mathrm{imize}{U}_{\mathrm{IA}}\{\left(W_{l,i}^{\left(a\right)}\right),\left({\mathrm{\Φ}}_k\right)\},\∀k,i,l$

$s.t.W_{l,i}^{\left(a\right)H}{W}_{l,i}^{\left(a\right)}=I_{L_l},l\∈\{1,...,N\}$

${\mathrm{\Φ}}_k^H{\mathrm{\Φ}}_k=I_{L_k},k\∈\{1,...,N\}$

Wherein $U_{\mathrm{IA}}\{\left(W_{l,i}^{\left(a\right)}\right),\left({\mathrm{\Φ}}_k\right)\}=\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\left|\right|{\mathrm{\Φ}}_k^H\left(\underset{l=1,l\≠k}{\overset{N}{\mathrm{\Σ}}}\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{H}_{k,i}{W}_{l,i}^{\left(\left(a\right)\right)}\right)\left|\right|}_F^2;$

Optimization aim represents and minimizes the summation that each user leaks into the interference of other users.

The implementation of described step 1) is as follows:

A () is first according to each receiving matrix Φ of optimization aim initialization _k, Φ _krepresent the receiving matrix of a kth user;

B () optimizes the first step pre-coding matrix of each user one by one in each base station optimization aim is rewritten as:

$W_{k,i}^{\left(a\right)}=\arg \underset{{\mathrm{\Φ}}_k^H{\mathrm{\Φ}}_k=I_{L_k},\∀k}{\min }\mathrm{tr}\left(W_{k,i}^{\left(a\right)H}\left(\underset{l=1,l\≠k}{\overset{N}{\mathrm{\Σ}}}{H}_{l,i}^H{\mathrm{\Φ}}_l{\mathrm{\Φ}}_l^H{H}_{l,i}\right)W_{k,i}^{\left(a\right)}\right);$

In solving result by matrix the L that several minimum characteristic values are corresponding _kcharacteristic vector forms, and wherein tr representing matrix asks mark computing;

C () be the receiving matrix Φ of optimizing user one by one in each base station _k, optimization aim is rewritten as

${\mathrm{\Φ}}_k=\arg \underset{W_{l,i}^{\left(a\right)H}{W}_{l,i}^{\left(a\right)}=I_{L_l},\∀l,i}{\min }\mathrm{tr}\left({\mathrm{\Φ}}_k^H\left(\underset{l=1,l\≠k}{\overset{N}{\mathrm{\Σ}}}\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{H}_{k,i}{W}_{l,i}^{\left(a\right)}{W}_{l,i}^{\left(a\right)H}{H}_{k,i}^H\right){\mathrm{\Φ}}_k\right);$

Φ in solving result _kby matrix the L that several minimum characteristic values are corresponding _kindividual characteristic vector composition;

D () repetitive process (b), (c), until the target function after adjacent twice iteration difference be less than the threshold value δ that presets or reach maximum iteration time regulation.

Described when solving receiving matrix in step (c), i-th base station can precompute value, and using this value as equivalent CSI in cooperative cluster share.

Described after optimizing the first step pre-coding matrix of each user and receiving matrix, step 2) implementation as follows:

A () is based on the first step pre-coding matrix of a kth user useful channel matrix H _k,iwith receiving matrix Φ _k, generate the equivalent channel matrix D of i-th base station to a kth user _k,i:

$D_{k,i}={\mathrm{\Φ}}_k^H{H}_{k,i}{W}_{k,i}^{\left(a\right)},\∀k,i$

D _k,irepresent the equivalent channel matrix of i-th base station to a kth user;

B () is according to equivalent channel matrix D _k,iwith signal vector s _krelevant information, calculate the phase rotation coefficient of each data flow of each user in base station end

For reducing the complexity that phase rotating process brings, need the value according to actual channel conditions determination phase rotation coefficient.Therefore equivalent channel matrix D is defined _k,ithe scale factor of middle element

${\mathrm{\μ}}_{p,q}^{(k,i)}=\frac{\left|d_{p,q}^{(k,i)}\mathrm{conj}\left(d_{p,q}^{(k,i)}\right)\right|}{{\mathrm{\Σ}}_{r=1}^{L_k}\left|d_{p,r}^{(k,i)}\mathrm{conj}(d_{p,r}^{(k,i)}\right|},p,q\∈\{1,...,L_k\}$

Wherein represent D _k,ithe element of p capable q row, represent D _k,ithe element of p capable r row, with represent s respectively _kq and p element, conj represents and gets conjugate operation, || the computing of expression delivery value.The threshold value δ that p is capable is set, will compare with δ:

When ${\mathrm{\&mu;}}_{p,q}^{(k,i)}<\mathrm{\&delta;}$ Time

When ${\mathrm{\&mu;}}_{p,q}^{(k,i)}\&GreaterEqual;\mathrm{\&delta;}$ Time

representative data relative to the phase rotating done;

Wherein represent D _k,ithe element of p capable q row, with represent s respectively _kq and p element, conj represents and gets conjugate operation, representative data relative to the phase rotating done;

C () calculates the phase rotation matrix for each user in each base station

represent the phase rotation matrix between i-th base station and a kth user;

D () utilizes minimum mean square error criterion, obtain the MMSE generator matrix M of i-th base station to a kth user _k,i:

$M_{k,i}=D_{k,i}^H{(D_{k,i}{D}_{k,i}^H+\frac{L_k{N}_0}{P_t^{(k,i)}}{I}_{L_k})}^{-1},\&ForAll;k,i$

Wherein N ₀represent the power spectral density of additive Gaussian noise, represent the transmitting power of i-th base station to a kth user, symbol _-1representing matrix inversion operation;

E () calculates the second step pre-coding matrix of i-th base station to a kth user wherein $W_{k,i}^{\left(b\right)}=M_{k,i}{R}_{k,i}^{\mathrm{\&phi;}};$

F () calculates final pre-coding matrix W _k,i, w _k,irepresent the pre-coding matrix of i-th base station to a kth user.

The described precoding processing through two steps, the signal vector r received by a kth user _kbe expressed as:

$r_k=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{f}_{k,i}{\mathrm{\&Phi;}}_k^H(H_{k,i}{W}_{k,i}^{\left(a\right)}{W}_{k,i}^{\left(b\right)}{s}_k+n_k),\&ForAll;k;$

Wherein f _k,irepresent the power contorl parameters of i-th base station to a kth user, and according to ${\left|\right|W_{k,i}^{\left(a\right)}{W}_{k,i}^{\left(b\right)}\left|\right|}_F^2=P_t^{(k,i)},$ R _kp element be written as:

$r_p^{\left(k\right)}\stackrel{\&CenterDot;}{=}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}(f_{k,i}\left(\underset{q\&Element;S}{\mathrm{\&Sigma;}}{\left|d_{p,q}^{(k,i)}\right|}^2{\left|s_q^{\left(k\right)}\right|}^2\right)s_p^{\left(k\right)}+{\tilde{n}}_p^{\left(k\right)});$

Wherein symbol represent approximately equal, for p element of the noise vector after receiver process.S represents a set, and in this set, q meets

Compared with prior art, the present invention has following useful effect:

The two step method for precoding of coordinate multipoint joint transmission system provided by the invention under multi-user scene, first by interference registration process, the optimization aim based on minimizing the interference of each user to other each users and leaking is set up at single base station end, and adopt alternating minimization method, obtain the first step pre-coding matrix for each user and receiving matrix respectively, make each use can effectively utilize system space resource to suppress inter-user interference per family, and receive glitch-free useful signal; Secondly, the receiving matrix for each user first step obtained, first step pre-coding matrix and useful channel matrix are combined into equivalent channel matrix, and corresponding phase rotation matrix is set on this basis, interference between each user's different antennae is converted into useful signal, effectively utilize each base station to the transmitting power of this user, and obtain second step pre-coding matrix and final pre-coding matrix.

Present invention utilizes the AF panel that interference alignment principle has been carried out between user, and be provided with corresponding phase rotation coefficient on this basis, interference between the data flow of user is converted into useful signal.The present invention does not need to be particularly limited to the system parameters such as the method for salary distribution of system antenna configuration, number of users and data flow, effectively make use of the transmitting power of space resources and base station.The error rate of the present invention declines obvious, accelerate the downward trend of ber curve, and system Mean Speed is also significantly improved.

Because institute of the present invention extracting method has carried out effective process to interference, in system antenna configuration and can utilize space resources in limited scene, its error performance and transmission rate are all better than traditional RBD method for precoding and based on the maximized method for precoding of SLNR.

Accompanying drawing illustrates:

Fig. 1 is system configuration schematic diagram.

Fig. 2 is user's average error rate and signal to noise ratio curve chart.

Fig. 3 is user's average transmission rate and signal to noise ratio curve chart, and wherein TSPCS represents the method for invention.

Embodiment:

The present invention will be described in detail below, and the explanation of the invention is not limited.

In the two step method for precoding that the present invention proposes:

The first step, according to interference alignment principle, inter-user interference overlap suffered by each user is placed in the orthogonal intersection space receiving space, use alternating minimization method at the first step pre-coding matrix of each user of base station end iteration optimization and receiving matrix, enable user receive glitch-free useful signal;

Second step, on the basis that inter-user interference suppresses, arranges suitable phase rotation coefficient, interference between the data flow suffered by each user is converted into useful signal, effectively utilizes base station transmitting power to improve systematic function.

Consider single cooperative cluster scene of multi-user joint transmission in Synergistic multi-point system, have M base station collaboration (M is positive integer), there is M i-th base station _itransmit antennas (i ∈ 1 ..., M}); N number of user (N is positive integer) is served in base station simultaneously, and a kth user is equipped with N _kroot antenna (k ∈ 1 ..., N}); The signal transmission fluxion L of a kth user _kbe not more than its reception antenna number, namely and the transmitting data stream of all users sum is no more than the antenna number of single base station, namely i-th base station is H to the channel matrix of a kth user _k,i(N _k× M _idimension), the received signal vector of a kth user is s _k(L _kdimension);

Adopt bulk nanometer materials model, channel matrix H _k,ifull rank and each element is separate.Assuming that different user data information is separate, namely k ≠ l and meet ( represent expectation computing, expression exponent number is L _kunit matrix).

The Received signal strength of user k for:

${\tilde{y}}_k={\mathrm{\&Phi;}}_k^H(\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{H}_{k,i}{W}_{k,i}{s}_k+\underset{l\&NotEqual;k,l=1}{\overset{N}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{H}_{k,i}{W}_{l,i}{s}_l+n_k)---\left(1\right)$

Wherein: (symbol [] _rcthe capable c column element of any r of representing matrix, symbol represent the multiple Gaussian Profile of Cyclic Symmetry); W _k,i(M _i× L _kdimension) be the pre-coding matrix of base station i to user k, n _kfor the additive white Gaussian noise of receiver k, and (N ₀for noise power spectral density); Φ _kfor the receiving matrix of user k, wherein symbol H represents conjugate transpose; Transmitting power uniform distribution.

The technological means of the two step method for precoding proposed is as follows:

Step 1 passes through interference registration process, by overlapping for the interference orthogonal intersection space being placed on reception space, realizes the reception of inter-user interference suppression and state no interference signal.

The target of interference registration process can be expressed as

${\mathrm{\&Phi;}}_k^H{H}_{k,i}{W}_{l,i}^{\left(a\right)}=0,\&ForAll;l\&NotEqual;k,i---\left(2\right)$

$\mathrm{rank}\left({\mathrm{\&Phi;}}_k^H{H}_{k,i}{W}_{k,i}^{\left(a\right)}\right)=L_k,\&ForAll;k,i---\left(3\right)$

Wherein represent the first step pre-coding matrix of i-th base station required by this step to a kth user, symbol rank represents acquisition rank of matrix.This formula represents in overlapping for the inter-user interference suffered by each user orthogonal intersection space being placed on reception space, enables user receive glitch-free useful signal.

When the base station in cooperative cluster and number of users more time, directly try to achieve and meet all users and disturb the closed solutions of aligned condition very difficult, therefore the present invention adopts alternating minimization method to obtain the suboptimal solution of formula (2) and (3).According to the interference aligned condition given by (2) formula and (3) formula, set up following optimization aim:

$\min \mathrm{imize}{U}_{\mathrm{IA}}\{\left(W_{l,i}^{\left(a\right)}\right),\left({\mathrm{\&Phi;}}_k\right)\},\&ForAll;k,i,l---\left(4\right)$

$s.t.W_{l,i}^{\left(a\right)H}{W}_{l,i}^{\left(a\right)}=I_{L_l},l\&Element;\{1,...,N\}---\left(5\right)$

${\mathrm{\&Phi;}}_k^H{\mathrm{\&Phi;}}_k=I_{L_k},k\&Element;\{1,...,N\}---\left(6\right)$

Wherein $U_{\mathrm{IA}}\{\left(W_{l,i}^{\left(a\right)}\right),\left({\mathrm{\&Phi;}}_k\right)\}=\underset{k=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left|\right|{\mathrm{\&Phi;}}_k^H\left(\underset{l=1,l\&NotEqual;k}{\overset{N}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{H}_{k,i}{W}_{l,i}^{\left(\left(a\right)\right)}\right)\left|\right|}_F^2;$

This optimization aim represents and minimizes the summation that each user leaks into the interference of other users.

Alternating minimization principle is, when there is N number of optimized variable, first temporarily can be fixed by N-1 variable, only optimizing some variablees, then changes the variable of optimization and temporarily fixing variable successively.After all variablees are all optimized, iteration carries out lower first step optimization, until acquired results meets given stopping criterion for iteration.Accordingly, the detailed process of this step execution is as follows:

A () is first according to each receiving matrix Φ of optimization aim initialization _k,

B () optimizes first step pre-coding matrix optimization aim is rewritten as by application matrix related properties

$W_{k,i}^{\left(a\right)}=\arg \underset{{\mathrm{\&Phi;}}_k^H{\mathrm{\&Phi;}}_k=I_{L_k},\&ForAll;k}{\min }\mathrm{tr}\left(W_{k,i}^{\left(a\right)H}\left(\underset{l=1,l\&NotEqual;k}{\overset{N}{\mathrm{\&Sigma;}}}{H}_{l,i}^H{\mathrm{\&Phi;}}_l{\mathrm{\&Phi;}}_l^H{H}_{l,i}\right)W_{k,i}^{\left(a\right)}\right)---\left(7\right)$

Solving result: by matrix the L that several minimum characteristic values are corresponding _kcharacteristic vector forms.Wherein symbol tr representing matrix asks mark computing.

C () optimizes receiving matrix Φ _k, optimization aim can be rewritten as

${\mathrm{\&Phi;}}_k=\arg \underset{W_{l,i}^{\left(a\right)H}{W}_{l,i}^{\left(a\right)}=I_{L_l},\&ForAll;l,i}{\min }\mathrm{tr}\left({\mathrm{\&Phi;}}_k^H\left(\underset{l=1,l\&NotEqual;k}{\overset{N}{\mathrm{\&Sigma;}}}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{H}_{k,i}{W}_{l,i}^{\left(a\right)}{W}_{l,i}^{\left(a\right)H}{H}_{k,i}^H\right){\mathrm{\&Phi;}}_k\right)---\left(8\right)$

Solving result: Φ _kby matrix the L that several minimum characteristic values are corresponding _kindividual characteristic vector composition.

D () repetitive process (b), (c), until the target function after adjacent twice iteration difference be less than the threshold value δ that presets or reach maximum iteration time regulation.

It may be noted that when solving receiving matrix, i-th base station can precompute value, and using this value as equivalent CSI information (channel condition information) in cooperative cluster share, information interaction expense can be reduced like this, and need not center processing unit be introduced.

Step 2, on the basis of step 1, obtains the equivalent channel matrix for different user in each base station, and arranges suitable phase rotation coefficient, and interference between data flow is converted into useful signal.1st step interference registration process is only disturbed user and suppresses, and underuses useful channel information raising receptivity.

2nd step is then on the basis of the 1st step, realizes the conversion disturbed between data flow.Detailed process is as follows:

A () is based on the first step pre-coding matrix of a kth user of step 1 gained useful channel matrix H _k,iwith receiving matrix Φ _k, generate the equivalent channel matrix D of i-th base station to a kth user _k,i:

$D_{k,i}={\mathrm{\&Phi;}}_k^H{H}_{k,i}{W}_{k,i}^{\left(a\right)},\&ForAll;k,i---\left(9\right)$

B () is according to equivalent channel matrix D _k,iwith signal vector s _krelevant information, generate phase rotation coefficient for reducing the complexity that phase rotating process brings, need the value according to actual channel conditions determination phase rotation coefficient.Therefore equivalent channel matrix D is defined _k,ithe scale factor of middle element

${\mathrm{\&mu;}}_{p,q}^{(k,i)}=\frac{\left|d_{p,q}^{(k,i)}\mathrm{conj}\left(d_{p,q}^{(k,i)}\right)\right|}{{\mathrm{\&Sigma;}}_{r=1}^{L_k}\left|d_{p,r}^{(k,i)}\mathrm{conj}(d_{p,r}^{(k,i)}\right|},p,q\&Element;\{1,...,L_k\}---\left(10\right)$

Wherein represent D _k,ithe element of p capable q row, represent D _k,ithe element of p capable r row, with represent s respectively _kq and p element, conj represents and gets conjugate operation, || the computing of expression delivery value.The threshold value δ that p is capable is set, will compare with δ:

When ${\mathrm{\&mu;}}_{p,q}^{(k,i)}<\mathrm{\&delta;}$ Time

When time

representative data relative to the phase rotating done;

C () calculates the phase rotation matrix between i-th base station and a kth user

D () utilizes least mean-square error (minimummeansquareerror, MMSE) criterion, and obtain the MMSE generator matrix M of i-th base station to a kth user _k,i

$M_{k,i}=D_{k,i}^H{(D_{k,i}{D}_{k,i}^H+\frac{L_k{N}_0}{P_t^{(k,i)}}{I}_{L_k})}^{-1},\&ForAll;k,i---\left(12\right)$

Wherein represent the transmitting power of i-th base station to a kth user, symbol _-1representing matrix inversion operation;

E () calculates the second step pre-coding matrix needed in i-th base station of this step acquisition to a kth user wherein $W_{k,i}^{\left(b\right)}=M_{k,i}{R}_{k,i}^{\mathrm{\&phi;}};$

F () calculates final pre-coding matrix W _k,i, w _k,irepresent the pre-coding matrix of i-th base station to a kth user.

Through the precoding processing of above two steps, the signal vector r received by a kth user _kcan be expressed as

$r_k=\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}{f}_{k,i}{\mathrm{\&Phi;}}_k^H(H_{k,i}{W}_{k,i}^{\left(a\right)}{W}_{k,i}^{\left(b\right)}{s}_k+n_k),\&ForAll;k---\left(13\right)$

Wherein f _k,irepresent the power contorl parameters of i-th base station to a kth user, according to this relation obtains.R _kp element can be written as

$r_p^{\left(k\right)}\stackrel{\&CenterDot;}{=}\underset{i=1}{\overset{M}{\mathrm{\&Sigma;}}}(f_{k,i}\left(\underset{q\&Element;S}{\mathrm{\&Sigma;}}{\left|d_{p,q}^{(k,i)}\right|}^2{\left|s_q^{\left(k\right)}\right|}^2\right)s_p^{\left(k\right)}+{\tilde{n}}_p^{\left(k\right)})---\left(14\right)$

Wherein symbol represent approximately equal, for p element of the noise vector after receiver process.S represents a set, and in this set, q meets this shows, after the method for precoding process that the present invention proposes, the interference between data flow is converted in order to useful signal.

Signal vector r is received user _kafterwards, for different modulation systems, need to adopt different demodulation modes:

(1) if s _kadopt PSK modulation.The demodulation of this modulation system does not need the amplitude information utilizing modulation symbol.Therefore the Received signal strength that (14) formula obtains directly can be carried out demodulation, and obtain useful signal information.

(2) if s _kadopt QAM modulation.This modulation system needs the amplitude information utilizing modulation symbol carry out demodulation.For reducing the expense that amplitude information is transmitted to user by base station, the mean value of the amplitude of the modulation symbol in a Frame can be obtained in base station end, this value is transferred to user as the amplitude information that demodulation is used.

Provide emulation and the Contrast on effect of said method below:

Consider three base station four user collaboration scenes in simulations, single base station transmit antennas number is 8, single-user receiver antenna number is 4, every user transmits the example that fluxion is 2, by the inventive method (TSPCS represents invention method therefor) and conventional wave beam forming (beamforming, BF) method, RBD method and SLNR maximization approach compare, to represent the effect that the present invention can reach, as shown in Figures 2 and 3.In this example, adopt QPSK modulation system (qam mode result is similar, just needs extra modulation symbol amplitude information), BF method, in RBD method and SLNR maximization approach, adopt zero-forcing receiver to receive data.

As seen from Figure 2, along with the increase of signal to noise ratio, interference becomes the principal element of influential system performance gradually.Traditional BF method does not suppress the process disturbed, so its error rate is higher, and declines slowly.And the reception antenna number that in the given scene of example, user is total is the twice of single base station transmit antennas number, make RBD method and the available space resources of SLNR maximization approach limited, error performance is sharply declined.And institute of the present invention extracting method is owing to processing the interference between inter-user interference and single user data flow well, therefore the error rate declines rapidly, shows good performance.Can find out, when threshold value δ >0, the performance of suggesting plans can not be decreased significantly simultaneously, but it is 2 owing to transmitting fluxion that the computation complexity of its phase rotating reduces 1/2(nearly).

In addition as seen from Figure 3, BF method, RBD method and SLNR maximization approach are due to the impact of residual interference, and its transmission rate is all lower, and growth trend is mild; And the transmission rate rapid development of institute of the present invention extracting method, when signal to noise ratio is 40dB, its value is more than 20bps/Hz, far above other three kinds of method for precoding.In addition, after introducing scale factor, the transmission rate of suggesting plans is substantially constant, and visible when threshold value is greater than 0, the judgement of the passing ratio factor carries out reduction to the computation complexity of phase rotating process can not affect transmission rate.

Claims (2)

1. the two step method for precoding of coordinate multipoint joint transmission system under multi-user scene, is characterized in that, comprise the following steps:

1) by interference registration process, inter-user interference overlap suffered by each user is placed in the orthogonal intersection space receiving space, and use alternating minimization method at the first step pre-coding matrix of each user of base station end iteration optimization and receiving matrix, enable user receive glitch-free useful signal;

2) on the basis that inter-user interference suppresses, each base station obtains the equivalent channel matrix for different user, phase rotation coefficient is set, interference between the data flow suffered by each user is converted into useful signal, effectively utilize base station transmitting power to improve systematic function;

In described interference registration process, the object representation of interference registration process is:

${\mathrm{\&Phi;}}_k^H{H}_{k,i}{W}_{l,i}^{\left(a\right)}=0,\&ForAll;l\&NotEqual;k,i;$

$rank\left({\mathrm{\&Phi;}}_k^H{H}_{k,i}{W}_{k,i}^{\left(a\right)}\right)=L_k,\&ForAll;k,i;$

Wherein represent the first step pre-coding matrix of i-th base station to a kth user, rank represents acquisition rank of matrix; Φ _kfor the receiving matrix of user k, H represents conjugate transpose; H _k,ibe the channel matrix of i-th base station to a kth user, L _kfor the transmitting data stream number of a kth user;

Adopt alternating minimization method to obtain the suboptimal solution of the target disturbing registration process, set up following optimization aim:

$\min imize{U}_{IA}\{\left(W_{l,i}^{\left(a\right)}\right),\left({\mathrm{\&Phi;}}_k\right)\},\&ForAll;k,i,l$

$s.t.W_{l,i}^{\left(a\right)H}{W}_{l,i}^{\left(a\right)}=I_{L_l},l\&Element;\{1,...,N\}$

${\mathrm{\&Phi;}}_k^H{\mathrm{\&Phi;}}_k=I_{L_k},k\&Element;\{1,...,N\}$

Wherein $U_{IA}\left\{\left(W_{l,i}^{\left(a\right)}\right),\left({\mathrm{\&Phi;}}_k\right)\right\}=\underset{k=1}{\overset{N}{\&Sigma;}}{||{\mathrm{\&Phi;}}_k^H\left(\underset{l=1,l\&NotEqual;k}{\overset{N}{\&Sigma;}}\underset{i=1}{\overset{M}{\&Sigma;}}{H}_{k,i}{W}_{l,i}^{\left(a\right)}\right)||}_F^2;$

Optimization aim represents and minimizes the summation that each user leaks into the interference of other users;

Described step 1) implementation as follows:

A () is first according to each receiving matrix Φ of optimization aim initialization _k, Φ _krepresent the receiving matrix of a kth user;

B () optimizes the first step pre-coding matrix of each user one by one in each base station optimization aim is rewritten as:

$W_{k,i}^{\left(a\right)}=\arg \underset{{\mathrm{\&Phi;}}_k^H{\mathrm{\&Phi;}}_k=I_{L_k},\&ForAll;k}{\min }tr\left(W_{k,i}^{\left(a\right)H}\left(\underset{l=1,l\&NotEqual;k}{\overset{N}{\&Sigma;}}{H}_{l,i}^H{\mathrm{\&Phi;}}_l{\mathrm{\&Phi;}}_l^H{H}_{l,i}\right)W_{k,i}^{\left(a\right)}\right);$

Wherein, argmin (A) represents that A can make this formula minimize, in solving result by matrix the L that several minimum characteristic values are corresponding _kcharacteristic vector forms, and wherein tr representing matrix asks mark computing;

C () be the receiving matrix of optimizing user one by one in each base station optimization aim is rewritten as

${\mathrm{\&Phi;}}_k=\arg \underset{W_{l,i}^{\left(a\right)H}{W}_{l,i}^{\left(a\right)}=I_{L_l},\&ForAll;l,i}{\min }tr\left({\mathrm{\&Phi;}}_k^H\left(\underset{l=1,l\&NotEqual;k}{\overset{N}{\&Sigma;}}\underset{i=1}{\overset{M}{\&Sigma;}}{H}_{k,i}{W}_{l,i}^{\left(a\right)}{W}_{l,i}^{\left(a\right)H}{H}_{k,i}^H\right){\mathrm{\&Phi;}}_k\right);$

Φ in solving result _kby matrix the L that several minimum characteristic values are corresponding _kindividual characteristic vector composition;

D () repetitive process (b), (c), until the target function after adjacent twice iteration difference be less than the threshold value η that presets or reach maximum iteration time regulation;

When solving receiving matrix in step (c), i-th base station can precompute value, and using this value as equivalent channel state information in cooperative cluster share;

After optimizing the first step pre-coding matrix of each user and receiving matrix, step 2) implementation as follows:

A () is based on the first step pre-coding matrix of a kth user useful channel matrix H _k,iwith receiving matrix Φ _k, generate the equivalent channel matrix D of i-th base station to a kth user _k,i:

$D_{k,i}={\mathrm{\&Phi;}}_k^H{H}_{k,i}{W}_{k,i}^{\left(a\right)},\&ForAll;k,i$

D _k,irepresent the equivalent channel matrix of i-th base station to a kth user;

B () is according to equivalent channel matrix D _k,iwith signal vector s _krelevant information, calculate the phase rotation coefficient of each data flow of each user in base station end for reducing the complexity that phase rotating process brings, need the value according to actual channel conditions determination phase rotation coefficient; Definition equivalent channel matrix D _k,ithe scale factor of middle element

${\mathrm{\&mu;}}_{p,q}^{(k,i)}=\frac{\left|d_{p,q}^{(k,i)}conj\left(d_{p,q}^{(k,i)}\right)\right|}{{\mathrm{\&Sigma;}}_{r=1}^{L_k}\left|d_{p,r}^{(k,i)}conj\left(d_{p,r}^{(k,i)}\right)\right|},p,q\&Element;\{1,...,L_k\}$

Wherein represent D _k,ithe element of p capable q row, represent D _k,ithe element of p capable r row, with represent s respectively _kq and p element, conj represents and gets conjugate operation, || the computing of expression delivery value; The threshold value δ that p is capable is set, will compare with δ:

When ${\mathrm{\&mu;}}_{p,q}^{(k,i)}<\mathrm{\&delta;}$ Time

When ${\mathrm{\&mu;}}_{p,q}^{(k,i)}\&GreaterEqual;\mathrm{\&delta;}$ Time

representative data relative to the phase rotating done;

C () calculates the phase rotation matrix for each user in each base station

represent the phase rotation matrix between i-th base station and a kth user;

D () utilizes minimum mean square error criterion, obtain the MMSE generator matrix M of i-th base station to a kth user _k,i:

$M_{k,i}=D_{k,i}^H{(D_{k,i}{D}_{k,i}^H+\frac{L_k{N}_0}{P_t^{(k,i)}}{I}_{L_k})}^{-1},\&ForAll;k,i$

Wherein N ₀represent the power spectral density of additive Gaussian noise, represent the transmitting power of i-th base station to a kth user, symbol _-1representing matrix inversion operation;

E () calculates the second step pre-coding matrix of i-th base station to a kth user wherein $W_{k,i}^{\left(b\right)}=M_{k,i}{R}_{k,i}^{\mathrm{\&phi;}};$

F () calculates final pre-coding matrix W _k,i, w _k,irepresent the pre-coding matrix of i-th base station to a kth user.

2. the two step method for precoding of coordinate multipoint joint transmission system under multi-user scene as claimed in claim 1, is characterized in that, through the precoding processing of two steps, and the signal vector r received by a kth user _kbe expressed as:

$r_k=\underset{i=1}{\overset{M}{\&Sigma;}}{f}_{k,i}{\mathrm{\&Phi;}}_k^H(H_{k,i}{W}_{k,i}^{\left(a\right)}{W}_{k,i}^{\left(b\right)}{s}_k+n_k),\&ForAll;k;$

Wherein f _k,irepresent the power contorl parameters of i-th base station to a kth user, and according to ${||W_{k,i}^{\left(a\right)}{W}_{k,i}^{\left(b\right)}||}_F^2=P_t^{(k,i)},$ R _kp element be written as:

Wherein symbol represents approximately equal, || represent the computing of delivery value, for p element of the noise vector after receiver process; S represents a set, and in this set, q meets