Background technology
MU-MIMO pre-coding system based on LTE-A, due to the increase of user side reception antenna number, often needs very large feedback overhead just can make the acquisition channel state information matrix of making a start, and then completes precoding.Existingly to the quantification of channel state information matrix, generally can be divided into two classes: one for to carry out by vector quantization the vector of channel state information matrix; Two for directly to quantize each element of channel state information matrix.As can be seen here, the feedback overhead to channel state information matrix, i.e. the size of quantification amount, not only depends on the size of the number of antennas Nt that makes a start, also depends on the size of reception antenna number N r.When in system, user antenna number N r is larger, though use feedback overhead less by vector quantization, also can produce larger feedback quantity.
Exist at present a kind of to match channels matrix R=H
hh carries out the feedback scheme of uniform quantization, and wherein H is channel state information matrix.This quantization scheme has utilized the size of R matrix only relevant to Nt, and with the irrelevant characteristic of Nr, and this match channels matrix R is the feedback overhead that the characteristic of conjugation symmetrical matrix reduces user side multiaerial system.
Particularly, above-mentioned that match channels matrix is carried out to the method concrete steps of uniform quantization is as follows, with the match channels matrix R to any one user k
kcarrying out uniform quantization is that example describes:
1) match channels matrix R
kdivided by H
kf norm, obtain
2) for matrix
element on diagonal, is all that arithmetic number and interval are positioned at [0,1], it is carried out to Q bit quantization in interval in [0,1] as follows:
Get
Will
arbitrary element on middle diagonal
successively to each a
nupper mapping, and therefrom determine minimum one, obtain
,
be quantified as
and for
can be by
represent.
3) for matrix
in lower triangle element, think that its real part and imaginary part are positioned at interval [1/2,1/2], can carry out respectively Q bit quantization as follows:
Get
Will
in arbitrary lower triangle element
real part and imaginary part respectively successively to b
nupper mapping, therefrom selects minimum one, obtains
,
be quantified as
utilize the conjugate symmetry of match channels matrix, the upper triangle element of its correspondence
be quantified as
By above-mentioned processing, can realize matrix
uniform quantization.User side is transmitted to base station by quantized result, and matrix is determined according to the quantized result receiving in base station
again according to each user's matrix
carry out the precoding of downlink data.
Above-mentioned method of match channels matrix R being carried out to uniform quantization is very applicable when Nt=2.But when Nt > 2, it is large that quantization error can become along with the increase of Nt.Make a concrete analysis of as follows:
Appoint the match channels matrix of taking family k
For arbitrary lower triangle element R
ij, i > j
According to basic inequality, for R
ij, i > jreal part and imaginary part have
Wherein, h
, r=real (h
.), h
, i=imag (h
.).
By (9), (10), can be found out, the first step in aforementioned uniform quantization method is got
Will make,
in the mould value of off-diagonal element when Nt > 2, along with the increase of Nt, diminish, appoint and get its lower triangle element
for example
, from (9), (10)
Wherein, trace (R
k) be the summation of all diagonal entries.By (13), (14), can be found out, when Nt > 2, along with the increase of Nt,
to diminish gradually, and in aforesaid uniform quantization method still in interval
right
quantize.So just cause quantization error can constantly become along with the increase of Nt large.
Summary of the invention
The invention provides a kind of uniform quantization method of match channels matrix, can be along with number of transmit antennas object increases and increase quantization error.
For achieving the above object, the present invention adopts following technical scheme:
A uniform quantization method for matrix, comprising:
Match channels matrix R to arbitrary user
koff diagonal element be normalized, structural matrix
off diagonal element
In interval [1/2,1/2], to matrix
off diagonal element carry out uniform quantization;
To described match channels matrix R
kdiagonal entry carry out uniform quantization, as matrix
the quantized result of middle diagonal entry.
Preferably, described to match channels matrix R
kdiagonal entry carry out uniform quantization and be:
Determine match channels matrix R
kdiagonal entry obey the card side that the degree of freedom is 2Nr and distribute, wherein, Nr is the reception antenna number of described user k;
According to predefined probability, the probability-distribution function that the card side that is 2Nr according to the described degree of freedom distributes, determines described match channels matrix R
kthe maximum of diagonal entry;
By described match channels matrix R
keach diagonal entry divided by described maximum, obtain normalized diagonal entry;
Described normalized diagonal entry is carried out to uniform quantization.
Preferably, describedly the off diagonal element of matrix carried out to uniform quantization be:
The lower triangle element of matrix is carried out to the uniform quantization in interval [1/2,1/2];
Uniform quantization result to described lower triangle element is asked conjugation, obtains with respect to diagonal being much of the uniform quantization result of triangle element;
Or,
The upper triangle element of matrix is carried out to the uniform quantization in interval [1/2,1/2];
Uniform quantization result to described upper triangle element is asked conjugation, obtains the uniform quantization result with respect to the lower triangle element of diagonal symmetry.
As seen from the above technical solution, in the present invention, when user's match channels matrix is carried out to uniform quantization, first off diagonal element is normalized,
in interval [1/2,1/2], to off diagonal element
carry out uniform quantization; By above-mentioned processing, the uniform quantization of off diagonal element in match channels matrix is converted to
uniform quantization, from formula (9) (10),
interval in [1/2,1/2], and its interval can not change with number of transmit antennas object, therefore, it carried out to the problem that uniform quantization also there will not be quantization error to increase with number of transmit antennas object.Next, then to the diagonal entry of match channels matrix carry out uniform quantization.
Embodiment
For making object of the present invention, technological means and advantage clearer, below in conjunction with accompanying drawing, the present invention is described in further details.
Basic thought of the present invention is: when match channels matrix is normalized, after consideration normalization, the interval problem of element, does not change its interval with number of transmit antennas object.
Below specific implementation of the present invention is described in detail.Uniform quantization method of the present invention comprises:
Step 101, is normalized structural matrix to the off diagonal element of user's match channels matrix
off diagonal element.
When carrying out the normalized of match channels matrix, guarantee for different number of transmit antennas, the interval of off diagonal element is fixed.In the present invention, based on this consideration, when carrying out off diagonal element normalization, according to the interval of aforementioned elements, determine normalized mode.Concrete, the off diagonal element after normalization is
The result of deriving in aforementioned background art
With
Visible, matrix
's
The imaginary part of off diagonal element and the interval of real part are all [1/2,1/2], and irrelevant with number of transmit antennas Nt.
Step 102, in interval [1/2,1/2], to matrix
off diagonal element carry out uniform quantization.
In this step, that carries out providing in the mode of uniform quantization and background technology is right
off diagonal element to carry out the mode of uniform quantization identical.Concrete, following triangle element is example, its real part and imaginary part are positioned at interval [1/2,1/2], can carry out respectively Q bits and quantize as follows:
Get
Will
in lower triangle element
real part and imaginary part respectively to b
nupper mapping, obtains
,
be quantified as
utilize the conjugate symmetry of match channels matrix, the upper triangle element of its correspondence
be quantified as
Certainly, also can first to upper triangle element, carry out uniform quantization, recycling conjugate symmetry obtains the uniform quantization result of lower triangle element.Or, also can directly to upper triangle element and lower triangle element, carry out independently of one another uniform quantization respectively, but this mode is more consuming time and waste resource.
Carry out by the way the uniform quantization of off diagonal element on match channels matrix, can avoid the problem that off diagonal element interval diminishes with the increase of number of transmit antennas object, thereby avoid the increase of quantization error.
Step 103, carries out uniform quantization to the diagonal entry of match channels matrix, as matrix
the quantized result of middle diagonal entry.
In this step, the diagonal entry uniform quantization mode of match channels matrix can adopt existing mode to carry out, according in background technology 1) and 2) mode process.
But in fact, although matrix
diagonal entry be all positioned in [0,1], but the actual interval of this diagonal entry in [0, x], and x < 1.In this case, in a larger interval, quantize, also can cause the increase of quantization error.
Based on above-mentioned consideration, in the present invention, provide a kind of quantification manner of diagonal entry.
Because the element in every user's channel matrix H can be approximated to be independent identically distributed multiple Gaussian random variable, obey CN (0,1).Therefore, the diagonal entry of its match channels matrix R is obeyed card side's distribution that the degree of freedom is 2Nr.The probability-distribution function distributing according to card side, for element R arbitrarily on diagonal in R
iican define a confidential interval
P(R
ii≤x)>d (20)
Like this, corresponding to a larger probability d, we can obtain an approximate R
iithe maximum x that may get.
Based on this, the mode of diagonal entry being carried out to uniform quantization can be:
First, determine match channels matrix R
kdiagonal entry obey the card side that the degree of freedom is 2Nr and distribute, wherein, Nr is the reception antenna number of user k; Then, according to predefined probability d, the probability-distribution function that the card side that is 2Nr according to the degree of freedom distributes, determines match channels matrix R
kthe maximum of diagonal entry; Next, by match channels matrix R
keach diagonal entry divided by maximum, obtain normalized diagonal entry; Normalized diagonal entry is carried out to uniform quantization.
For example, table 1 has provided the table that card side distributes, and during for Nr=2, while getting d=0.9, can obtain x ≈ 8 by table 1.Therefore, R
ii/ x can be similar to interval [0,1] and quantize, and ought carry out Q
dduring the quantification of bit, have
Get
Will
element on middle diagonal is to c
nupper mapping, obtains
a
0.995 0.99 0.975 0.95 0.9 0.1 0.05 0.025 0.01 0.005
df
1 0.00004 0.00016 0.001 0.004 0.016 2.706 3.841 5.024 6.635 7.879
2 0.01 0.02 0.051 0.103 0.211 4.605 5.991 7.378 9.21 10.597
3 0.072 0.115 0.216 0.352 0.584 6.251 7.815 9.348 11.345 12.838
4 0.207 0.297 0.484 0.711 1.064 7.779 9.488 11.143 13.277 14.86
5 0.412 0.554 0.831 1.145 1.61 9.236 11.07 12.833 15.086 16.75
6 0.676 0.872 1.237 1.635 2.204 10.645 12.592 14.449 16.812 18.548
7 0.989 1.239 1.69 2.167 2.833 12.017 14.067 16.013 18.475 20.278
8 1.344 1.646 2.18 2.733 3.49 13.362 15.507 17.535 20.09 21.955
9 1.735 2.088 2.7 3.325 4.168 14.684 16.919 19.023 21.666 23.589
10 2.156 2.558 3.247 3.94 4.865 15.987 18.307 20.483 23.209 25.188
11 2.603 3.053 3.816 4.575 5.578 17.275 19.675 21.922 4.725 26.757
12 3.074 3.571 4.404 5.226 6.304 18.549 21.026 23.337 26.217 28.3
13 3.565 4.107 5.009 5.892 7.042 19.812 22.362 24.736 27.688 29.819
14 4.075 4.66 5.629 6.571 7.79 21.064 23.685 26.119 29.141 31.319
15 4.601 5.229 6.262 7.261 8.547 22.307 24.996 27.488 30.578 32.801
16 5.142 5.812 6.908 7.962 9.312 23.542 26.296 28.845 32 34.267
Table 1
Can carry out uniform quantization to the diagonal entry of match channels matrix by the way, and using this uniform quantization result as matrix
the quantized result of diagonal entry.And, in this quantification manner, for diagonal entry, determine interval more accurately, utilize this interval to quantize, reduced quantization error.
By abovementioned steps 101-103, can realize the uniform quantization for match channels matrix.User side can send to base station by this quantized result, and base station receives after quantized result, according to the quantized result of diagonal entry, can recover the diagonal entry utilizing after x normalization, and then is multiplied by x and obtains the diagonal entry R before normalization
ii.Next, utilize each diagonal entry recovering and the off diagonal element after normalization.Like this, just obtain match channels matrix, just can utilize this matrix to carry out precoding.
By the processing of the invention described above, the technique effect that can be achieved as follows: for the element on the off-diagonal of match channels matrix R, the uniform quantization scheme proposing has avoided existing uniform quantization scheme when Nt > 2, and its quantization error becomes large this defect along with the increase of Nt; Further, for the element on R diagonal of a matrix, the relatively little diagonal entry of usining replaces R matrix trace as quantizing object, has effectively reduced the quantization error quantizing under big or small condition waiting.
Quantization algorithm in the solution of the present invention and background technology is carried out to Performance Ratio, and simulation result is shown in Fig. 1 and Fig. 2.Concrete simulated conditions is respectively in Table 2 and table 3.
In Fig. 1, abscissa represents the number of transmitting antenna, and ordinate represents the mean square error of quantized result.Wherein, curve 101 represents the uniform quantization method in background technology, carries out 3bit quantification; Curve 102 represents the uniform quantization method in the present invention, and off diagonal element and diagonal entry all carry out 3bit quantification.In Fig. 1, can find out, for the system of Nr=2, along with the increase of Nt number, uniform quantization scheme of the present invention is compared to existing uniform quantization scheme, can reduce greatly the generation of quantization error.
In Fig. 2, abscissa represents signal to noise ratio, and ordinate represents the error rate.Wherein, curve 201 represents not consider that quantization error situation carries out the performance curve after precoding; The performance curve of curve 202 for utilizing the method for background technology to carry out carrying out again in uniform quantization situation precoding, adopts 3 bit quantizations during quantification; Curve 203 is for to utilize method of the present invention to carry out carrying out in uniform quantization situation the performance curve of precoding again, and during quantification, diagonal and off diagonal element are 3 bit quantizations.As seen from Figure 2, in SLNR algorithm, thereby utilize uniform quantization result of the present invention to carry out precoding, compare by existing scheme at Ber=10
-3time, can improve the gain more than 2dB.
Parameter |
Value |
Base station transmit antennas number |
[2∶1∶10] |
User's reception antenna number |
2 |
Channel type |
Flat Rayleigh fading channel |
X value |
|
8 |
Quantization parameter Q |
3bits |
Quantization parameter Qd |
3bits |
Table 2
Parameter |
Value |
Base station transmit antennas number |
4 |
User's reception antenna number |
2 |
Channel type |
Flat Rayleigh fading channel |
Signal to noise ratio |
[0:2:10]dB |
X value |
|
8 |
Quantization parameter Q |
3bits |
Quantization parameter Qd |
3bits |
Table 3
The foregoing is only preferred embodiment of the present invention, in order to limit the present invention, within the spirit and principles in the present invention not all, any modification of making, be equal to replacement, improvement etc., within all should being included in the scope of protection of the invention.