CN102412939A - Method for carrying out channel information feedback by using double codebook and system thereof - Google Patents

Abstract

The invention provides a method for carrying out channel information feedback by using double codebook. The method comprises the following steps that: a signal receiver selects a grade-one optimal code word from codebook of a precoding matrix in the prior art; the signal receiver combines channel state information and a function formed by the grade-one optimal code word and selects a grade-two optimal code word from preset grade-two codebook according to a capacity maximum standard; the signal receiver generates feedback information comprising a code word index of the selected grade-one optimal code word and the code word index of the grade-two optimal code word and sends to a signal sender; the signal sender generates the precoding matrix which is a product of the grade-one optimal code word and the grade-two optimal code word according to the received feedback information. The invention also provides a system of carrying out the channel information feedback by using the double codebook. Compared to the channel information feedback method in the prior art, by using the method, the channel information can be fed back accurately.

Description

A kind of method and system that adopts dicode originally to carry out feedback of channel information

Technical field

The present invention relates to multiple-input and multiple-output (MIMO) technology in the wireless communication technology field, relate in particular to a kind of method and system that adopts dicode originally to carry out feedback of channel information.

Background technology

Precoding technique is the channel information according to receive direction transmit leg feedback; Data to sending are handled with pre-coding matrix; The equivalence in the space channel transmission course of each data flow is transmitted for separate channel, reached through the balanced purpose that receives signal of a plurality of reception antennas.

In LTE R8, for two antenna opening transmission, promptly { 0,1}, code book number are 2 * 2=4 to antenna opening p ∈, and pre-coding matrix W (i) will select from table 1 or subclass wherein.To the Closed-Loop Spatial Multiplexing pattern that defines among the LTE R8, when number of plies υ=2, can not use code book index 0.

Table 1

For four antenna opening transmission, promptly { 0,1,2,3}, code book number are 4 * 4=16 to p ∈, and pre-coding matrix W will select from table 2 or subclass wherein.

Expression is from expression formula

Choose { matrix that s} row constitute, I is 4 * 4 unit matrix in the formula, u _nValue is seen table 2.

Table 2

Can find out that from table 1 and table 2 precoding codebook of the prior art is unitary matrice, the mould perseverance of unitary matrice is 1, so pre-coding matrix multiply by and transmit, and can not change the mould that transmits.

Want to select best pre-coding matrix rightly, just must understand the current channel condition of transmitter.Channel condition is provided by the feedback information of the MIMO receiver of creating closed-loop system.For the downlink transmission of LTE precoding, portable terminal or subscriber equipment (UE) measured channel characteristic information, and definite pre-coding matrix index (PMI), CQI (CQI) and/or rank index feedback informations such as (RI); Said feedback information sends to base station (eNB), selects precoding codebook by the base station according to said feedback information.In the prior art because the base station can only select single precoding codebook, and this single precoding codebook the channel condition with reality is corresponding well, cause the feedback of channel information accurate inadequately.

Summary of the invention

The invention provides a kind of method and system that adopts dicode originally to carry out feedback of channel information, can be than single precoding codebook feedback channel information more accurately.

The embodiment of the invention proposes a kind of method that adopts dicode originally to carry out feedback of channel information, comprises the steps:

A, receiving side signal are selected a code word Fp as the optimum code word of one-level from the code book of the pre-coding matrix of prior art;

The function that B, receiving side signal combine channel condition information and the optimum code word of said one-level to form, the standard that maximizes is by measure selected the optimum code word F of secondary from the secondary code book that is provided with in advance; The optimum code word F of said secondary makes matrix for approaching diagonal matrix, and wherein H is the channel matrix of signalling to receiving side signal;

C, receiving side signal generate the feedback information of the codewords indexes of the codewords indexes that comprises the optimum code word of selected one-level and the optimum code word of secondary, and said feedback information is sent to signalling;

D, signalling are according to the codewords indexes that comprises in institute's receiving feedback information; One-level code book that inquiry is provided with in advance and secondary code book obtain optimum code word of corresponding one-level and the optimum code word of secondary, and generating pre-coding matrix is the product of optimum code word of one-level and the optimum code word of secondary.

Preferably; Code word in the said secondary code book is the approximate value of

; Wherein, θ satisfies with

:

Preferably, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

Preferably, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and said secondary code book is selected θ=[π/8 for use; 3 π/8; 5 π/8; 7 π/8; 9 π/8; 11 π/8; 13 π/8; 15 π/8] corresponding functional value, and cos in the secondary code book (θ) and sin (θ) used exwponential function approximation: as θ during near 2k π, cos (θ) ≈ exp (j θ) and sin (θ) ≈ 0; And as θ during, sin (θ) ≈ exp (j θ+j π) and cos (θ) ≈ 0 near pi/2+2k π.

Preferably, said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

Preferably, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and receiving side signal adopts 4 antenna ports, and channel sequence is 1, and said secondary code book is got θ=[2 π/8; 4 π/8; 6 π/8; 8 π/8; 10 π/8; 12 π/8; 14 π/8; 16 π/8] time corresponding exponential function value.

Preferably, said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

The codewords indexes of second level code book	Pre-coding matrix
		0	F(:，:，1)＝0.7071+0.7071i
1	F(:，:，2)＝0.0000+1.0000i
		2	F(:，:，3)＝-0.7071+0.7071i
3	F(:，:，4)＝-1.0000+0.0000i
		4	F(:，:，5)＝-0.7071-0.7071i
5	F(:，:，6)＝-0.0000-1.0000i
		6	F(:，:，7)＝0.7071-0.7071i
7	F(:，:，8)＝1.0000-0.0000i

The embodiment of the invention also proposes a kind of system that adopts dicode originally to carry out feedback of channel information, comprises receiving side signal and signalling,

Said receiving side signal is selected a code word F from the code book of the pre-coding matrix of prior art _pAs the optimum code word of one-level; In conjunction with the function of channel condition information and the optimum code word composition of said one-level, the standard that maximizes is by measure selected the optimum code word F of secondary from the secondary code book that is provided with in advance; The optimum code word F of said secondary makes matrix For approaching diagonal matrix, wherein H is the channel matrix of signalling to receiving side signal; And, generate the codewords indexes comprise the optimum code word of selected one-level and the feedback information of the codewords indexes of secondary optimum code word, and said feedback information is sent to signalling;

Said signalling is used for the codewords indexes that comprises according to institute's receiving feedback information; One-level code book that inquiry is provided with in advance and secondary code book obtain optimum code word of corresponding one-level and the optimum code word of secondary, and generating pre-coding matrix is the product of optimum code word of one-level and the optimum code word of secondary.

Preferably; Code word in the said secondary code book is the approximate value of

; Wherein, θ satisfies with

:

Preferably, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

Preferably, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and said secondary code book is selected θ=[π/8 for use; 3 π/8; 5 π/8; 7 π/8; 9 π/8; 11 π/8; 13 π/8; 15 π/8] corresponding functional value, and cos in the secondary code book (θ) and sin (θ) used exwponential function approximation: as θ during near 2k π, cos (θ) ≈ exp (j θ) and sin (θ) ≈ 0; And as θ during, sin (θ) ≈ exp (j θ+j π) and cos (θ) ≈ 0 near pi/2+2k π.

Preferably, said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

Preferably, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and receiving side signal adopts 4 antenna ports, and channel sequence is 1, and said secondary code book is got θ=[2 π/8; 4 π/8; 6 π/8; 8 π/8; 10 π/8; 12 π/8; 14 π/8; 16 π/8] time corresponding exponential function value.

Preferably, said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

The codewords indexes of second level code book	Pre-coding matrix
		0	F(:，:，1)＝0.7071+0.7071i
1	F(:，:，2)＝0.0000+1.0000i
		2	F(:，:，3)＝-0.7071+0.7071i
3	F(:，:，4)＝-1.0000+0.0000i
		4	F(:，:，5)＝-0.7071-0.7071i
5	F(:，:，6)＝-0.0000-1.0000i
		6	F(:，:，7)＝0.7071-0.7071i
7	F(:，:，8)＝1.0000-0.0000i

Can find out from above technical scheme, adopt dicode originally to carry out feedback of channel information, can originally more accurately reflect actual channel condition than solid size.The embodiment of the invention gives the structure foundation of second level code book, and the structural scheme of a plurality of second level code book.Those skilled in the art carry out various modification and are equal to replacement the concrete form of the disclosed second level of embodiment code book under the guidance of inventive concept, all should be included in protection scope of the present invention.

Description of drawings

Fig. 1 carries out Monte Carlo simulation, the sketch map of the f that obtains (θ) and

cumulative distribution function for the independent same distribution Rayleigh channel under the scene that 22 are received;

The employing dicode that Fig. 2 proposes for the embodiment of the invention originally carries out the flow chart of feedback of channel information;

Value that Fig. 3 is

and signal to noise ratio concern sketch map;

Fig. 4 is the sketch map that the simulation result of dicode this programme and prior art solid size this programme compares;

Fig. 5 is 2 situation for the second level code book shown in the table 4 being used for 4 antenna ports and channel sequence, the comparison sketch map of simulation result that obtains and prior art;

Fig. 6 is 1 situation for the second level code book shown in the table 5 being used for 4 antenna ports and channel sequence, the comparison sketch map of simulation result that obtains and prior art.

Fig. 7 is 1 situation for the second level code book shown in the table 5 being used for 2 antenna ports and channel sequence, the comparison sketch map of simulation result that obtains and prior art.

Embodiment

Correlative study shows that based on the code book scheme of unitary matrice precoding, Grassmannian code book scheme can obtain bigger performance gain in the contrast prior art.But the Grassmannian precoding is a kind of non-unitary matrice precoding, does not have constant modulus property.

Consider one and be furnished with M _tTransmit antennas, M _rThe point to point wireless communication system of root reception antenna.The at first multiplexing M≤min (M that becomes of input signal _t, M _r) individual different sub stream s, through pre-coding matrix F _pCarry out preliminary treatment and obtain actual transmitting, receiving terminal equivalence baseband signal expression formula is following:

y＝HF _ps+v (1)

Wherein, H represents channel matrix, and v represents noise vector, and the independent identically distributed multiple normal distribution CN of the element obedience of channel matrix (0, N ₀), the covariance matrix of noise vector is N ₀I _M

When receiving terminal adopted linear receiver such as least mean-square error (MMSE) receiver, channel capacity was calculated as follows:

$r(H,F_p)=\underset{k=1}{\overset{M}{\mathrm{\Σ}}}\mathrm{Lo}{g}_2(1+{\mathrm{SNR}}_k(H,F_p))$

$=-\mathrm{Lo}{\mathrm{<figure-callout\; id="2"\; label="g"\; filenames="HSA00000281039600022.png,HSA00000281039600051.png"\; state="\{\{state\}\}">g</figure-callout>}}_2{\mathrm{MN}}_0-{\mathrm{<figure-callout\; id="2"\; label="Log"\; filenames="HSA00000281039600022.png,HSA00000281039600051.png"\; state="\{\{state\}\}">Log</figure-callout>}}_2\underset{k=1}{\overset{M}{\mathrm{\&Pi;}}}{\left[{(F_p^H{H}^H{\mathrm{HF}}_p+{\mathrm{MN}}_0{I}_M)}^{-1}\right]}_{\mathrm{kk}}---\left(2\right)$

By the Hadmard inequality, have

$\underset{k=1}{\overset{M}{\mathrm{\&Pi;}}}{\left[{(H_p^H{H}^H{\mathrm{HF}}_p+{\mathrm{MN}}_0{I}_M)}^{-1}\right]}_{\mathrm{kk}}\&GreaterEqual;\det \left[{(H_p^H{H}^H{\mathrm{HF}}_p+{\mathrm{MN}}_0{I}_M)}^{-1}\right]---\left(3\right)$

So obtain the channel capacity upper bound do

$r^{\mathrm{upper}}(H,F_p)=-\mathrm{Lo}{g}_{2}M{N}_0-\mathrm{<figure-callout\; id="2"\; label="Lo\; g"\; filenames="HSA00000281039600022.png,HSA00000281039600051.png"\; state="\{\{state\}\}">Lo</figure-callout>}{g}_{}{}_2\det \left[{(F_p^H{H}^H{\mathrm{HF}}_P+{\mathrm{MN}}_0{I}_M)}^{-1}\right]---\left(4\right)$

And if only if for channel capacity

Obtain when deteriorating to diagonal matrix, at this moment

V wherein _HObtain through channel H being carried out singular value decomposition (SVD), even D wherein _HThe diagonal matrix that constitutes for the singular value of channel matrix H.

When adopting dicode the design, as establish former pre-coding matrix F _pAs the one-level code book, second level code book is F, then should construct F and make Be (approaching) diagonal matrix.

When channel sequence is 2, known by matrix theory: spin matrix can reach the diagonalizable effect of arbitrary symmetrical matrix.Therefore, can design form be the second level code book F of spin matrix, expression formula is following:

For making

approach diagonal matrix, the element in the forced coding matrix of this moment satisfies:

Wherein $\left[\begin{array}{cc}{a}_{11}& a_{12}\\ a_{21}& a_{22}\end{array}\right]=F_p^H{H}^H{\mathrm{HF}}_P---\left(7\right)$

Be the expression f (θ) that tries one's best complete with limited codeword set, the possible value of

; The cumulative distribution function (CDF) that needs research f (θ),

; Independent same distribution Rayleigh channel under the scene of 22 receipts (2 transmitting antennas and 2 reception antennas) is carried out Monte Carlo simulation; Obtain result as shown in Figure 1; Wherein circle is represented the CDF of f (θ), and triangle is represented the CDF of

.Can know by result shown in Figure 1; The non-homogeneous growth of CDF of f (θ),

; So can select the value set of secondary encoder matrix F according to its CDF, make it concentrate on the intensive value district of f (θ),

.

It is as shown in Figure 2 that the employing dicode that the embodiment of the invention proposes originally carries out the flow process of feedback of channel information, comprises the steps:

Step 201: the code word selection scheme that the receiving side signal utilization is traditional such as capacity criterion, singular value criterion etc. are selected a code book as the optimum code word of one-level from the code book of the pre-coding matrix of prior art;

Step 202: the function that the code word that receiving side signal combines the channel condition information and the first step to select is formed maximization principle is by measure selected the optimum code word of secondary from the secondary code book that is provided with in advance.Promptly

$F=\arg \min \underset{k=1}{\overset{M}{\mathrm{\&Sigma;}}}{\left[{(F^H{F}_p^H{H}^H{\mathrm{HF}}_{P}F+N_0{I}_M)}^{-1}\right]}_{\mathrm{kk}}.$

Step 203: receiving side signal generates the codewords indexes comprise the optimum code word of selected one-level and the feedback information of the codewords indexes of secondary optimum code word, and said feedback information is sent to signalling.

Step 204: signalling is according to the codewords indexes that comprises in institute's receiving feedback information; One-level code book that inquiry is provided with in advance and secondary code book obtain optimum code word of corresponding one-level and the optimum code word of secondary; Generating pre-coding matrix is the product of optimum code word of one-level and the optimum code word of secondary, promptly

W＝F _p·F。

In general, said receiving side signal is a subscriber equipment, and signalling is the base station.

Above dicode the design is used its influence to channel capacity of emulation when single user's multiple-input and multiple-output (SU-MIMO) situation.The emulation situation is 22 to be received, and channel circumstance is independent identically distributed flat Rayleigh fading channel.

First order code book uses the code book among the LTE R8; Second level code book is considered the CDF distribution of f (θ),

and sets that second level code book is 8 pre-coding matrixes or its subclass as shown in table 3:

Table 3

The selection of first order code book and second level code book is all pressed channel capacity maximization criterion and is selected.Receiving terminal uses least mean-square error (MMSE) decoder.

It is pointed out that the code book that is used for 2 antenna ports among the LTE R8 has only two encoder matrixs when the transmission number of data layers is 2; Promptly only need 1 feedback bit; Second level code book in the embodiment of the invention has 8 pre-coding matrixes, so need 3 feedback bits, total feedback quantity is 4 bits.

Can find out by simulation result, this feedback of solid size among the contrast LTE R8, the performance of this feedback of dicode is keeping the gain of about 1bit/s/Hz always.

Value that is

shown in Figure 3 and signal to noise ratio concern sketch map.As can beappreciated from fig. 3; Than high s/n ratio the time; The value of

is very little; So when code book designs; Can disregard the influence of

; And only study the CDF of f (θ), other elements in this moment pre-coding matrix are monotonic function with f (θ) and shine upon, so its CDF is consistent with each other.

For guaranteeing the constant modulus property of code book, can consider the element in the code book is replaced with exponential function.Consider as θ during cos (θ) ≈ exp (j θ) and sin (θ) ≈ 0 near 2k π; And as θ during near pi/2+2k π, sin (θ) ≈ exp (j θ+j π) and cos (θ) ≈ 0 construct code book as shown in table 4 based on this, promptly select θ=[π/8 for use; 3 π/8; 5 π/8; 7 π/8; 9 π/8; 11 π/8; 13 π/8; 15 π/8], and respectively cos (θ) in the former secondary code book and sin (θ) are used exwponential function approximation.

Table 4

Performance simulation is seen Fig. 4.Wherein, rice font icon representation adopts the simulation result of the dicode this programme shown in the table 3, and the triangle icon is represented the simulation result of dicode this programme of second level code book shown in the employing table 4, and circular icon is represented the simulation result of solid size this programme of the prior art.Can know by Fig. 4,, guarantee the constant modulus property of code book though dicode this programme of second level code book shown in the employing table 4 slightly descends with respect to the performance of the dicode this programme shown in the employing table 3.

In like manner, it is 2 situation that the code book of the second level shown in the table 4 is used for 4 antenna port orders, can get simulation result Fig. 5, and rice font icon representation adopts the simulation result of the dicode this programme shown in the table 4, and circular icon is represented the simulation result of solid size this programme of the prior art.

In like manner, be 1 situation for 4 antenna ports and channel sequence, obey equally distributed rayleigh fading channel in the decline phase place, second level code book can directly be got θ=[2 π/8; 4 π/8; 6 π/8; 8 π/8; 10 π/8; 12 π/8; 14 π/8; 16 π/8] time corresponding exponential function value, second level code book form is as shown in table 5, can get simulation result Fig. 6.Rice font icon representation adopts the simulation result of the dicode this programme shown in the table 5, and circular icon is represented the simulation result of solid size this programme of the prior art.

The codewords indexes of second level code book	Pre-coding matrix
		0	F(:，:，1)＝0.7071+0.7071i
1	F(:，:，2)＝0.0000+1.0000i
		2	F(:，:，3)＝-0.7071+0.7071i
3	F(:，:，4)＝-1.0000+0.0000i
		4	F(:，:，5)＝-0.7071-0.7071i
5	F(:，:，6)＝-0.0000-1.0000i
		6	F(:，:，7)＝0.7071-0.7071i
7	F(:，:，8)＝1.0000-0.0000i

Table 5

In like manner, be 1 situation for 2 antenna port orders, second level code book is got the code book of the same table 5, can get analogous diagram 7.Rice font icon representation adopts the simulation result of the dicode this programme shown in the table 5, and circular icon is represented the simulation result of solid size this programme of the prior art.

Can find out all that from Fig. 5-7 simulation result that the dicode that the embodiment of the invention proposes originally carries out feedback of channel information all improves significantly than the simulation result that the solid size of prior art originally carries out feedback of channel information.

The above is merely preferred embodiment of the present invention, and is in order to restriction the present invention, not all within spirit of the present invention and principle, any modification of being made, is equal to replacement, improvement etc., all should be included within the scope that the present invention protects.

Claims (14)

1. a method that adopts dicode originally to carry out feedback of channel information is characterized in that, comprises the steps:

A, receiving side signal are selected a code word F from the code book of the pre-coding matrix of prior art _pAs the optimum code word of one-level;

The function that B, receiving side signal combine channel condition information and the optimum code word of said one-level to form, the standard that maximizes is by measure selected the optimum code word F of secondary from the secondary code book that is provided with in advance; The optimum code word F of said secondary makes matrix

for approaching diagonal matrix, and wherein H is the channel matrix of signalling to receiving side signal;

C, receiving side signal generate the feedback information of the codewords indexes of the codewords indexes that comprises the optimum code word of selected one-level and the optimum code word of secondary, and said feedback information is sent to signalling;

D, signalling are according to the codewords indexes that comprises in institute's receiving feedback information; One-level code book that inquiry is provided with in advance and secondary code book obtain optimum code word of corresponding one-level and the optimum code word of secondary, and generating pre-coding matrix is the product of optimum code word of one-level and the optimum code word of secondary.

2. method according to claim 1 is characterized in that, the code word in the said secondary code book does

Approximate value, wherein, θ with

Satisfy:

$\left[\begin{array}{cc}{a}_{11}& a_{12}\\ a_{21}& a_{22}\end{array}\right]=F_p^H{H}^H{\mathrm{HF}}_P.$

3. method according to claim 2 is characterized in that, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

4. method according to claim 2 is characterized in that, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and said secondary code book is selected θ=[π/8 for use; 3 π/8; 5 π/8; 7 π/8; 9 π/8; 11 π/8; 13 π/8; 15 π/8] corresponding functional value, and cos in the secondary code book (θ) and sin (θ) used exwponential function approximation: as θ during near 2k π, cos (θ) ≈ exp (j θ) and sin (θ) ≈ 0; And as θ during, sin (θ) ≈ exp (j θ+j π) and cos (θ) ≈ 0 near pi/2+2k π.

5. method according to claim 4 is characterized in that, said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

6. method according to claim 2 is characterized in that, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and receiving side signal adopts 4 antenna ports, and channel sequence is 1, and said secondary code book is got θ=[2 π/8; 4 π/8; 6 π/8; 8 π/8; 10 π/8; 12 π/8; 14 π/8; 16 π/8] time corresponding exponential function value.

7. method according to claim 6 is characterized in that, said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

The codewords indexes of second level code book Pre-coding matrix 0 F(:，:，1)＝0.7071+0.7071i 1 F(:，:，2)＝0.0000+1.0000i 2 F(:，:，3)＝-0.7071+0.7071i 3 F(:，:，4)＝-1.0000+0.0000i 4 F(:，:，5)＝-0.7071-0.7071i 5 F(:，:，6)＝-0.0000-1.0000i 6 F(:，:，7)＝0.7071-0.7071i 7 F(:，:，8)＝1.0000-0.0000i

8. a system that adopts dicode originally to carry out feedback of channel information comprises receiving side signal and signalling, it is characterized in that,

Said receiving side signal is selected a code word F from the code book of the pre-coding matrix of prior art _pAs the optimum code word of one-level; In conjunction with the function of channel condition information and the optimum code word composition of said one-level, the standard that maximizes is by measure selected the optimum code word F of secondary from the secondary code book that is provided with in advance; The optimum code word F of said secondary makes matrix

For approaching diagonal matrix, wherein H is the channel matrix of signalling to receiving side signal; And, generate the codewords indexes comprise the optimum code word of selected one-level and the feedback information of the codewords indexes of secondary optimum code word, and said feedback information is sent to signalling;

Said signalling is used for the codewords indexes that comprises according to institute's receiving feedback information; One-level code book that inquiry is provided with in advance and secondary code book obtain optimum code word of corresponding one-level and the optimum code word of secondary, and generating pre-coding matrix is the product of optimum code word of one-level and the optimum code word of secondary.

9. system according to claim 8 is characterized in that, the code word in the said secondary code book does Approximate value, wherein, θ with

Satisfy:

$\left[\begin{array}{cc}{a}_{11}& a_{12}\\ a_{21}& a_{22}\end{array}\right]=F_p^H{H}^H{\mathrm{HF}}_P.$

10. system according to claim 9 is characterized in that, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

11. system according to claim 9 is characterized in that, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and said secondary code book is selected θ=[π/8 for use; 3 π/8; 5 π/8; 7 π/8; 9 π/8; 11 π/8; 13 π/8; 15 π/8] corresponding functional value, and cos in the secondary code book (θ) and sin (θ) used exwponential function approximation: as θ during near 2k π, cos (θ) ≈ exp (j θ) and sin (θ) ≈ 0; And as θ during, sin (θ) ≈ exp (j θ+j π) and cos (θ) ≈ 0 near pi/2+2k π.

12. system according to claim 11 is characterized in that, said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

13. system according to claim 9 is characterized in that, the pre-coding matrix of said prior art is the code book of the regulation among the LTE R8, and receiving side signal adopts 4 antenna ports, and channel sequence is 1, and said secondary code book is got θ=[2 π/8; 4 π/8; 6 π/8; 8 π/8; 10 π/8; 12 π/8; 14 π/8; 16 π/8] time corresponding exponential function value.

14. system according to claim 13 is characterized in that, said secondary code book is set or its subclass that the pre-coding matrix shown in the following table is formed:

The codewords indexes of second level code book Pre-coding matrix 0 F(:，:，1)＝0.7071+0.7071i 1 F(:，:，2)＝0.0000+1.0000i 2 F(:，:，3)＝-0.7071+0.7071i 3 F(:，:，4)＝-1.0000+0.0000i 4 F(:，:，5)＝-0.7071-0.7071i 5 F(:，:，6)＝-0.0000-1.0000i 6 F(:，:，7)＝0.7071-0.7071i 7 F(:，:，8)＝1.0000-0.0000i

Images