CN101695008B - Multi-user multi-antenna two-stage limited feedback method - Google Patents

Abstract

A multi-user multi-antenna two-stage limited feedback method comprises the steps: a user end quantifies channel state information into channel quality information and obtains low-order channel direction information through small-size codebooks to implement first-stage feedback; a base station selects an optimal user group as a scheduling user, and notifies the scheduling user to feed back high-order channel direction information and uses the low-order channel direction information as a precoding matrix to begin sending data to the scheduling user; the scheduling user acquires the high-order channel direction information via large-size codebooks and implements second-stage feedback, and the base station computes in accordance with the high-order channel direction information which is fed back by the scheduling user to obtain the optimized precoding matrix, and sends data to the scheduling user in accordance with the optimized precoding matrix. The method has the characteristics of high transmission rate and low limited feedback, and specific implementation methods can be provided for the limited feedback policy of cellular cells in the third generation, beyond third generation and the fourth generation cellular mobile communication systems, etc.

Description

Multi-user multi-antenna two-stage limited feedback method

Technical field

What the present invention relates to is a kind of method of radio transmission technical field, specifically is a kind of multi-user multi-antenna two-stage limited feedback method.

Background technology

Many antennas (MIMO) system is through the multiple antenna of transmitting terminal and receiving terminal, and the channel capacity of wireless communication system can be improved.In the MIMO broadcast channel, the base station transmits information for a plurality of single antenna users through many antennas simultaneously.The transmission performance of multi-user MIMO system is relevant with channel condition information.When the base station can access complete channel information, in gross power one regularly, system's overall transmission rate increased and increases along with user's number, reaches multi-user diversity gain.Yet when the base station did not have channel condition information, systematic function and tdma system were suitable.The information of each user feedback is limited in the real system, and the base station can only receiving unit channel condition information, i.e. Limited Feedback system.Multiuser MIMO Limited Feedback scheme is the basis with the linear predictive coding, compares nonlinear precoding (for example optimum DPC technology), the linear predictive coding simplicity of design, and the transmission performance loss is little, more is prone to realize in the real system kind.

Linear pre-coding method is divided into orthogonal random beam shaping (ORBF) and close-to zero beam shaping (ZFBF).ORBF comprises the precoding vector group that generates at random and sends the maximum user of SINR respectively to.Therefore, each user need not to feed back self channel information, as long as feedback channel quality information (CQI): maximum SINR and channel directional information (CDI): the vector of corresponding optimum number.This scheme is widely used because step is simple.Feedback scheme based on the ORBF technology has 2 kinds: (1) quadrature space division multiplexing (OSDMA): each user one group of precoding vector group is at random passed in the base station, SINR and corresponding vector that user feedback is maximum.The theoretical proof system can access multi-user diversity.(2) selectivity quadrature space division multiplexing (OSDMA-S): the base station produces many group precoding vectors at every turn at random, the maximum Signal Interference and Noise Ratio (SINR) of user feedback with corresponding vectorial number, increased the limit feedback quantity but promoted transmission performance.In the reality, the space division multiplexing of selectivity quadrature is called PU2RC again, has become the 3GPP-LTE standard.When using ZFBF, the base station uses channel directional information, the channel quality information of feedback to confirm channel vector, is used to the pre-coding matrix that user's scheduling and design close-to zero beam are shaped, to eliminate inter-user interference.When there being the complete channel state information, inter-user interference can be eliminated fully, and its performance is near optimal case simultaneously.Yet, in order to make throughput of system continue to rise, avoid the ceiling benefit along with SNR increases, feedback quantity must be linear with signal noise ratio (SNR, the dB of unit).Feedback scheme based on ZFBF has 2 kinds: (1) method of exhaustion: choose the user group identical with number of transmit antennas at every turn; Through calculating the overall transmission rate that uses the close-to zero beam forming technique to reach, choose the maximum user's group of overall transmission rate at last and be the optimal user group.(2) compel zero quadrature space division multiplexing (ZF-OSDMA): after user CSI is confirmed in the base station; Choose and each selected user's coefficient correlation and the optimum user of channel gain associating through each iteration; In each iteration; At first from user's group to be selected, screen out and organize the user of coefficient correlation, therefrom choose again with selected user and organize the maximum user of quadrature component and try to achieve pre-coding matrix greater than threshold value with selected user.This method is considered between the user correlation and is repeatedly obtained optimal user group after the iteration.Use orthogonal random beam shaping (ORBF) and close-to zero beam shaping (ZFBF) to be used for providing the feedback resources of CSIT information only to use once, though can both realize multi-user diversity, both each defectiveness in scheduling and precoding vector design.

Complete Limited Feedback scheme is divided into user's scheduling and precoding two parts.The precoding vector of ORBF is confirmed in advance, but because vectorial number is few, accurately the user dispatches; The precoding vector of ZFBF is through calculating, and in order to reduce inter-user interference, each user must feed back more precise channels directional information, has increased feedback quantity.Therefore, be necessary to seek a kind of scheme of Limited Feedback more efficiently.

Literature search through to prior art is found; In " E-UTRA Precoding FeedbackOverhead " (E-UTRA precoding feedback mechanism) that Motorolar company delivers, think that the feedback resources amount that is used to dispatch with the precoding vector design demand is different on the 49th meeting of 3GPP TSG RAN WG1 (in May, 2007) motion R1-072175.Scheduling phase needs all user feedback channel informations, but each user's feedback quantity seldom; On the contrary, the precoding stage fraction user feedback that only need be scheduled precise channels information more.So, can feedback resources be divided into two stages fully and utilize respectively.

Summary of the invention

The present invention is directed to the above-mentioned deficiency that prior art exists; A kind of multi-user multi-antenna two-stage limited feedback method is provided; Pass through two-stage limited feedback method; Make the thick channel information that quantizes of each user feedback of phase I help base station user scheduling, the thin channel information that quantizes of user feedback that the second stage base station requires to be scheduled obtains precoding vector.The present invention has promoted transmission rate and low being easy to of feedback quantity realized.

The present invention realizes through following technical scheme, the present invention includes following steps:

Step 1: the base station pilot signal transmitted is given user side, and user side receives pilot signal and channel condition information;

Described channel condition information refers to channel vector, and each element in this channel vector is the channel fading coefficient of base station respective antenna to the user side antenna;

Step 2: user side is quantified as channel condition information channel quality information and obtains low order channel directional information through the small size code book, and carries out the phase I feedback, and channel quality information and low order channel directional information are sent to the base station;

Described channel quality information refers to by channel condition information and the letter that total transmission snr computation obtains and disturbs the ratio of making an uproar.

Described small size code book code word number is 2 ³-2 ⁴

Described low order channel directional information refers to the code word sequence number of mating most with channel vector in the small size code book, and wherein, low order channel directional information is used the small size code book, and channel quantitative is coarse, and its result identifies with low order channel directional information.

Step 3: user's scheduling is carried out by channel quality information, the low order channel directional information of user feedback in the base station; Choose optimal user group as dispatched users; And notice dispatched users feedback high-order channel directional information; And use low order channel directional information as pre-coding matrix, begin to send data to dispatched users;

Described user's scheduling refers to the base station and by low order channel directional information all users is sorted out the user who also therefrom chooses the channel quality information maximum, again with the user grouping of choosing, therefrom chooses channel quality information and maximum user's group;

Described optimal user group refers to user channel quality information and maximum user's group in the group;

Step 4: dispatched users obtains high-order channel directional information by the large scale code book; And carry out second stage and feed back; High-order channel directional information is sent to the base station; The base station calculates the optimization pre-coding matrix according to the high-order channel directional information of dispatched users feedback, and sends data to dispatched users according to optimizing pre-coding matrix;

Described large scale code book code word number is 2 ¹⁰-2 ¹⁸Individual;

Described optimization pre-coding matrix is meant: the pseudo inverse matrix of the high-order channel directional information of dispatched users.

The present invention is divided into two stages with feedback resources, and the phase I is used for user's scheduling, and second stage is used for precoding.Than existing program, this programme has reduced feedback quantity when effectively increasing the throughput of system.Therefore, the scheme that the present invention proposes has the characteristics of high transfer rate and low Limited Feedback, can for the third generation (3G), ultra three generations (S3G), the 4th generation system such as (4G) cellular mobile communication cellular cell Limited Feedback strategy concrete implementation method is provided.

Description of drawings

Fig. 1 is a sketch map of the present invention.

Fig. 2 is the sequential chart of embodiment.

Fig. 3 is code book orthogonal code group design example figure.

Fig. 4 is different Limited Feedback scheme transmission performance analogous diagram.

Embodiment

Elaborate in the face of embodiments of the invention down, present embodiment provided detailed execution mode and concrete operating process, but protection scope of the present invention is not limited to following embodiment being to implement under the prerequisite with technical scheme of the present invention.

Two stage Limited Feedback schemes in the multiuser MIMO downlink that present embodiment relates to, system block diagram is as shown in Figure 1, and whole system is by multi-antenna base station, single antenna user side, and unlimited channel, Limited Feedback channel are formed.Multi-antenna base station contains the M transmit antennas, and the base station receives the information of user side through the Limited Feedback channel feedback, carries out user's scheduling, and the waiting message of dispatched users is carried out precoding processing, is transmitted into wireless channel through antenna at last; K independent user receives the information that is transmitted by the base station from wireless channel, obtains the own part that needs, and carries out Channel Processing according to these information simultaneously, after the channel vector value is quantized through the Limited Feedback channel feedback to the base station.Wireless channel and Limited Feedback channel are separate at frequency domain.

The sequential of the whole process of present embodiment is as shown in Figure 3, specifically comprises the steps:

(1) the base station pilot signal transmitted is given user side

The base station pilot signal transmitted, base station and transmitting terminal are known during the pilot signal off-line, are used for Channel Processing.

(2) user side receives pilot signal, processing channel state information

Because user side is known the pilot signal that send the base station in advance, after the pilot signal of acceptance by Channel Transmission,, can obtain channel vector h through comparing and various channel processing method _k∈ ^{1 * M}, i.e. channel condition information.

(3) user side is quantified as channel quality information and low order channel directional information with channel condition information

User side uses the small size code book C based on the FFT matrix structure, its each code word c _nBe N * M FFT matrix delegation,

$C=\left[\begin{array}{c}{c}_1\\ {\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">c</figure-callout>}}_2\\ .\\ .\\ .\\ c_N\end{array}\right]=\frac{1}{\sqrt{M}}\left[\begin{array}{cccc}{\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">e</figure-callout>}}^j& {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">e</figure-callout>}}^j& ...& {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">e</figure-callout>}}^j\\ {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">e</figure-callout>}}^j& {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">e</figure-callout>}}^j& ...& {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">e</figure-callout>}}^j\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">e</figure-callout>}}^j& {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">e</figure-callout>}}^j& ...& {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">e</figure-callout>}}^j\end{array}\right]---\left(1\right)$

Formula makes u=[u in (1) ₁U _M] be undetermined coefficient.Any two different code word c _mAnd c _{M '}Coefficient correlation does between (m ≠ m ') | c _mc _{M '} ^H|.Through selecting suitable u, can make between code word coefficient correlation very little, thereby code word is evenly distributed on vector space based on the code book of FFT matrix structure.In the simulating, verifying process, u=when getting N=8 [1 23 5].

The low order channel directional information of user feedback be among the small size code book C with channel vector h _kThe code word d that cross-correlation coefficient is maximum _k,

$d_k=\arg \underset{c_n\&Element;C}{\max }|h_k\&CenterDot;c_n^H|/\left|\right|h_k\left|\right|---\left(2\right)$

The channel quality information q of user feedback _kFor user k can reach the transmission rate processing costs.Because it is quadrature that the user who is scheduled organizes corresponding low order channel directional information.Number of users when scheduling | S _Opt| these code words constitute one group of complete orthogonal basis of M dimensional vector during=M.User k ∈ S can reach transmission rate processing costs q _kSatisfy,

$q_k=\log (1+\frac{\mathrm{\&rho;}{\left|\right|h_k{d}_k\left|\right|}^2}{1+\mathrm{\&rho;}{\mathrm{\&Sigma;}}_{i\&Element;S_{\mathrm{opt}}\backslash \left\{k\right\}}{\left|\right|h_k{d}_i\left|\right|}^2})=\log (1+\frac{\mathrm{\&rho;}{\left|\right|h_k{d}_k\left|\right|}^2}{1+\mathrm{\&rho;}({\left|\right|h_k\left|\right|}^2-{\left|\right|h_k{d}_k\left|\right|}^2)})---\left(3\right)$

(4) client feeds back channel quality information and low order channel directional information are given the base station

The user by up Limited Feedback channel, feeds back to base station at time slot 1 with channel quality information and low order channel directional information.Up Limited Feedback channel is that frequency domain separates with the downlink data transmission channel.

(5) user's scheduling is carried out by the information of user feedback in the base station, chooses optimal user group

Because small size code book C is known, can obtains many groups in advance according to code book and have M the Codeword Sets of quadrature each other.Concrete grammar is: making the topological diagram of code book, is node with the code word, coefficient correlation when between any code word $\left|c_m{c}_{m^{\&prime;}}^H\right|=0,$ Be quadrature, line between two nodes.The last whole topological diagram of traversal search, arbitrarily M in twos all the node of line be one group of quasiorthogonal code word group S _r(r=1 ..., R).Fig. 3 is codebook size N=8, the code book topological diagram during u=[1 23 5], and wherein the quasiorthogonal code word is totally 4 groups, is respectively: { 1-2-5-6}, { 2-3-6-7}, { 3-4-7-8}, { 4-5-8-1}.

User's scheduling, at first, the base station is referred to the user quantization areas of corresponding codewords representative with reference to the low order channel directional information of user feedback.Make code word c _nRepresent in the quantization areas user's set to be G _nFor each user k, according to its low order channel directional information d _kValue be included into

Promptly

$G_n=\left\{k\right|\&ForAll;d_k=c_n,k=1...K\}---\left(4\right)$

Secondly, the quantization areas G that each code word is represented _nIn the quality information q that wins the confidence _kMaximum user screens out all the other users.Make the quantization areas G of each code word representative _nIn the user number chosen be v (n), then v (n) satisfies

$v\left(n\right)=\arg \underset{k\&Element;G_n}{\max }{q}_k---\left(5\right)$

At last, relatively in each Codeword Sets the code word respective user can reach overall transmission rate and processing costs be q _kThe user who is scheduling with maximum that group user of, value organize S _Opt,

$S_{\mathrm{opt}}=\underset{S_r,r=1...R}{\max }\underset{v\left(n\right),n\&Element;S_r}{\mathrm{\&Sigma;}}{q}_{v\left(n\right)}---\left(6\right)$

(6) the user feedback high-order channel directional information that is scheduled is notified in the base station, and begins to send data

S is notified user k ∈ in the base station _OptFeedback high-order channel directional information is used for accurate precoding design.

Simultaneously, because the low order channel directional information of dispatched users is crossing in twos, can be directly as pre-coding matrix.The base station temporarily utilizes this pre-coding matrix to come to transmit data for the user of scheduling.

(7) user who is scheduled obtains high-order channel directional information by the large scale code book

In the precoding stage, used the large scale code book C that quantizes (RVQ) based on random vector _RVQ, each code word all is that independent random produces in the code book, obeys the vector of multiple Gaussian distribution.It should be noted that and use any specific aim designed code book all than good based on the performance of random vector quantification code book.

The user uses large scale code book C _RVQ, the high-order channel directional information of feedback, D _kBe code book C _RVQIn with channel vector h _kThe code word that cross-correlation coefficient is maximum,

$D_k=\arg \underset{c_n\&Element;C_{\mathrm{RVQ}}}{\max }|h_k\&CenterDot;c_n^H|/\left|\right|h_k\left|\right|---\left(7\right)$

(8) the user feedback high-order channel directional information D that is scheduled _k

The user who is scheduled is with high-order channel directional information D _kBy up Limited Feedback channel, feed back to the base station at time slot 2.

(9) base station obtains more accurately pre-coding matrix according to the dispatched users feedback calculation

Order $S_{\mathrm{Opt}}=\{{\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">s</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="H2009101975154E00022.png"\; state="\{\{state\}\}">s</figure-callout>}}_2...s_{\left|S_{\mathrm{Opt}}\right|}\},$ $H={\left[\begin{array}{cccc}{D}_{s_1}^H& D_{s_2}^H& ...& D_{s_{\left|S_{\mathrm{Opt}}\right|}}^H\end{array}\right]}^H,$ Then pre-coding matrix W satisfies,

W＝H ^H(HH ^H) ^-1 (8)

Owing to used high-order channel directional information, W compares more accurate with the pre-coding matrix in (6), can eliminate inter-user interference better, increases throughput of system.

(10) transmission information in base station is given dispatched users

The base station sends the data of dispatched users to user by the M transmit antennas through down channel simultaneously after precoding processing.Instantly one-period begins the above step of repetition, ensures that the base station of wireless area and all users normally move.

Present embodiment is than existing program, and this programme has reduced feedback quantity when effectively increasing the throughput of system.These can be confirmed from emulation.

As shown in Figure 4, compared various emission strategies: the overall transmission rate of two stage Limited Feedback strategies, EU2RC.Wherein two stage Limited Feedback strategy phase I small size code book code word number N1 are 8 (using preceding 8 code words of 4 antenna N=16 code books of 3GPPTS 36.211 V8.30 Table 6.3.4.2.3-2), and the value of second stage large scale code book code word number N2 gets 2 ¹², 2 ¹⁴, 2 ¹⁸And inf (infinity, promptly the second rank channel directional information is a perfect condition, its value equals h _k/ || h _k||); EU2RC (using the 4 antenna N=16 code books of 3GPP TS 36.211 V8.30 Table6.3.4.2.3-2) code book code word number is 16.Whole system, transmitting terminal antenna number are 4, and the user side antenna number is 1, transmitting terminal signal to noise ratio P/N=10dB.

Along with number of users increases, the throughput of system of three kinds of strategies also increases, and has realized user diversity.Two stage Limited Feedback strategies are along with second stage code book code word number increases, and throughput constantly rises.Relatively 3 kinds of strategies use the throughput of system of two stage Limited Feedback strategies bigger.

Compare feedback quantity, each user feedback 3bit low order channel directional information of two-stage policy phase I, second stage only 4 users need be fed back 12-18bit high-order channel directional information; And each user of EU2RC will feed back 4bit channel directional information.So when number of users (surpassed 100) greatly, the total feedback quantity of two-stage policy was littler.

Therefore; Two stage Limited Feedback schemes are a kind of prioritization schemes in the multiuser MIMO downlink that present embodiment proposes; Effectively reduced feedback quantity the throughput of increase system the time, the multi-subscriber dispatching strategy that can be many antennas cellular cell provides important basis and concrete implementation method.

Claims (4)

1. a multi-user multi-antenna two-stage limited feedback method is characterized in that, may further comprise the steps:

Step 1: the base station pilot signal transmitted is given user side, and user side receives pilot signal and channel condition information;

Step 2: user side is quantified as channel condition information channel quality information and obtains low order channel directional information through the small size code book, and carries out the phase I feedback, and channel quality information and low order channel directional information are sent to the base station;

Step 3: user's scheduling is carried out by channel quality information, the low order channel directional information of user feedback in the base station; Choose optimal user group as dispatched users; And notice dispatched users feedback high-order channel directional information; And use low order channel directional information as pre-coding matrix, begin to send data to dispatched users;

Step 4: dispatched users obtains high-order channel directional information by the large scale code book; And carry out second stage and feed back; High-order channel directional information is sent to the base station; The base station calculates the optimization pre-coding matrix according to the high-order channel directional information of dispatched users feedback, and sends data to dispatched users according to optimizing pre-coding matrix;

Described channel condition information refers to channel vector, and each element in this channel vector is the channel fading coefficient of base station respective antenna to the user side antenna;

Described channel quality information refers to by channel condition information and the letter that total transmission snr computation obtains and disturbs the ratio of making an uproar;

Described low order channel directional information refers to the code word sequence number of mating most with channel vector in the small size code book, and low order channel directional information is used the small size code book, identifies with low order channel directional information;

Described user's scheduling refers to the base station and by low order channel directional information all users is sorted out the user who also therefrom chooses the channel quality information maximum, again with the user grouping of choosing, therefrom chooses channel quality information and maximum user's group;

Described optimization pre-coding matrix is meant: the pseudo inverse matrix of the high-order channel directional information of dispatched users.

2. multi-user multi-antenna two-stage limited feedback method according to claim 1 is characterized in that, described small size code book code word number is 2 ³-2 ⁴

3. multi-user multi-antenna two-stage limited feedback method according to claim 1 is characterized in that, described optimal user group refers to the maximum user's group of user channel quality information sum in the group.

4. multi-user multi-antenna two-stage limited feedback method according to claim 1 is characterized in that, described large scale code book code word number is 2 ¹⁰-2 ¹⁸Individual.

Images