WO2011097881A1

WO2011097881A1 - Method and system for multi-level feedback of channel information

Info

Publication number: WO2011097881A1
Application number: PCT/CN2010/077076
Authority: WO
Inventors: 陈艺戬; 李儒岳; 张峻峰; 李书鹏; 徐俊
Original assignee: 中兴通讯股份有限公司
Priority date: 2010-02-12
Filing date: 2010-09-17
Publication date: 2011-08-18
Also published as: CN101771514A; CN101771514B

Abstract

A method for multi-level feedback of channel information is provided in the present invention. The method includes: the receiver feeds back at least two-level channel indication information to the transmitter, and the transmitter acquires the channel information according to the channel indication information fed back by the receiver. A system for multi-level feedback of channel information is also provided in the present invention. By feeding back multi-level channel information, the present invention can improve the feedback precision of the channel information and reduce the calculation complexity.

Description

Method and system for multi-level feedback of channel information

Technical field

The present invention relates to the field of wireless communications, and in particular, to a multi-level feedback method and system for channel information. Background technique

In wireless communication, if multiple antennas are used by both the transmitting end and the receiving end, spatial multiplexing can be used to obtain a higher rate. Compared with the general spatial multiplexing method, an enhanced technique is that the receiving end feeds back channel information to the transmitting end, and the transmitting end uses some transmitting precoding techniques according to the obtained channel information, which greatly improves the transmission performance. A simple implementation method is to directly use channel feature vector information for precoding, which is mainly used in single-user MIMO (multiple input multiple output). There are other better but more complex methods, mainly used in multi-user MIMO, but all require more accurate channel information.

In the Long Term Evolution (LTE) system, the feedback of channel information mainly utilizes a simple single codebook feedback method, and the performance of MIMO transmit precoding technology mainly depends on the accuracy of the codebook feedback.

The codebook based channel information quantization feedback is based on the assumption that the limited feedback channel capacity is B bps/Hz. Then the number of available codewords is = 2 ³ . The eigenvector space of the channel matrix is quantized to form a codebook space 9 ΐ = ^, ^ _{2 The} transmitting end and the receiving end jointly save or generate the codebook in real time (same as the transceiver end). For each channel implementation, the receiving end selects a codeword that best matches the channel according to certain criteria, and feeds the codeword sequence number 1 back to the transmitting end. The transmitting end finds the precoding codeword according to the serial number, and obtains channel information, which is mainly feature vector information of the channel.

Generally, the 3⁄4 can be further divided into multiple codebooks corresponding to Rank, and each Rank corresponds to a plurality of codewords to quantize the precoding matrix formed by the channel feature vectors under the Rank. Since the number of Rank and non-zero eigenvectors of the channel are equal, in general, the codeword when ank is N will have N columns. So we can divide the codebook 9 into multiple subcodebooks according to Rank. 91

Number of layers _w (Rank)

1 2 N

%

Number of columns with 1 Number of columns 2 Number of columns with N Number of codewords Vector codeword matrix Matrix

The codewords to be stored in Rank>l are in the form of a matrix. For example, the codebook in the LTE protocol is a feedback method using such codebook quantization, as shown in Table 2. It is worth noting that the codebook in Table 2 It is a codebook that is most suitable for single-user MIMO with quantization accuracy and overhead.

Table 2 Precoding codebook of downlink 4 antenna

The LTE-Advanced Long Term Evolution Advanced (LTE-A) needs to support larger system bandwidth (up to 100MHz) and needs to improve the average spectral efficiency and the frequency error efficiency of cell edge users. Therefore, LTE-A needs to enhance some technologies, such as multi-user MIMO (MU-MIMO), which is an unacceptable task of the feedback accuracy of the existing LTE single-user codebook. In LTE, SU-MIMO and MU-MIMO share a set of codebooks. Since MU-MIM0 is limited to Rank1, it is stipulated that the feature vector is found only from its corresponding 4-bit codebook according to the codebook feedback method of Rank1. Quantify the codeword and give feedback. In LTE-A, since the codebook accuracy of the MU needs to be further enhanced, a more expensive codebook is required. At the same time, in order to obtain better dynamic switching gain, SU-MIMO and MU-MIMO are combined into the same transmission mode. There must be a feedback mode feedback technology that needs to support both SU-MIMO and MU-MIMO applications.

One method in the prior art re-designs a set of high-cost codebooks according to the required MU-MIMO codebook overhead, for example: using the 6-bit Rankl codebook in the 802.16m standard, as shown in Table 3.

Table 3 802.16m codebook

Index (Codebook)

0 0.5000 -0.5000 0.5000 -0.5000

1 -0.5000 -0.5000 0.5000 0.5000

2 -0.5000 0.5000 0.5000 -0.5000

3 0.5000 -0.5000I 0.5000 -0 5000i

4 -0.5000 -0.5000I 0.5000 0.5000i

5 -0.5000 0.5000i 0.5000 -0.5000i

6 0.5000 0.5000 0.5000 0.5000

7 0.5000 0.5000i 0.5000 0.5000i

8 0.5000 0.5000 0.5000 -0.5000

9 0.5000 0.5000i -0.5000 0.5000i

10 0.5000 -0.5000 0.5000 0.5000

11 0.5000 -0.5000i -0.5000 -0.5000i

12 0.5000 0.3536 + 0.3536i 0.5000i -0.3536 + 0.3536i /s/u/ O iz-z-ososld Ϊ88/-60Π0ΖAV

/s/u/ O iz-z-ososld Ϊ88/-60Π0ΖAV

63 0.3082 0.2254 - 0.22041 -0.4888 0.40761 -0.6302 + 0.05881 In Table 3, the rank of each row represents a Rankl codebook, for independent high-overhead codebooks such as higher overhead codebooks and high Rank codebooks. Design, the number of codewords grows exponentially, which is a very complicated technical problem, and the compatibility with the original codeword is not good.

The drawbacks of the prior art are: Poor compatibility with the LTE protocol. In the feedback, it is necessary to feed back the PMI information of the SU codebook, and it is necessary to simultaneously feed back the codebook indicating the MU. Assuming that the MU codebook is 6 bits, then the high Rank requires feedback of lObit codebook information. In addition, the receiving end UE needs to store a plurality of code words of multiple sets of codebooks, and performs multiple codebook selections when the codebook is selected, and the calculation amount is large.

Another method in the prior art uses the correlation matrix information of the channel to rotate the LTE codebook:

C = R ² xC _RS

Where C is the codebook obtained after the C _{RS is} rotated, R = H ^H H , H is the dimension xN channel matrix, and N is the number of receiving antennas, which is the number of transmitting antennas.

The distribution information of the feature vector is included in the channel correlation matrix information. After the codeword in the codebook is rotated, the distribution probability of the feature vector is also quantified in a region where the distribution probability is large, and the quantization precision is further improved.

However, this method can only solve the problem of quantization accuracy in a scene with certain correlation, and has no effect on non-related scenes, and the non-correlated channel scene is often a hot scene such as urban microcell. The feedback of the other channel correlation matrix also requires a large feedback overhead. Summary of the invention

The technical problem to be solved by the present invention is to provide a channel information hierarchical feedback method and system with small feedback overhead, good compatibility, flexible feedback combination, low computational complexity when codeword selection is low, and high overall quantization precision.

In order to solve the above problem, the present invention provides a method for multi-level feedback of channel information, including: receiving, by a receiving end, at least two levels of channel indication information to a transmitting end; The transmitting end acquires channel information according to the channel indication information fed back by the receiving end.

The method may further have the following feature: the step of the receiving end feeding back at least two levels of channel indication information to the sending end includes:

The receiving end feeds the M-level channel indication information to the transmitting end, where the M>1, the m-th channel indication information corresponds to a codeword of the m-th codebook^4, m=\..M, indicating the code An index of the codeword in the ^; wherein, the codeword of the codebook 4 of the first level can independently represent the channel information;

The M code words of the M-level codebook. . . , _3⁄4 } collectively indicate that one channel information is Ω ( _Α ..... A _MS , Ω is a function, and the transmitting end is fed back according to the receiving end. The step of the channel indication information acquiring channel information includes: the sending end acquiring channel information according to Ω(Α _ΙΑ ...^ _Μ ).

The method may also have the following features, the Ω being one of the following operations or a combination thereof: left multiplication, right multiplication, dot multiplication, difference, and sum operation.

The method may also have the following feature: the m + 1 level codebook _{+1 is} independent of the code word of the mth level codebook

The method may further have the following feature: the m+1 level codebook ₊₁ exists at a center, and the ₊₁ includes or does not include the center o _m+1 , and the ₊₁ The code word except ₁ in the center o _m+1 is the same as the distance from the center.

The receiving end feeds the M-level channel indication information to the transmitting end, where the M>1, the m-th channel indication information corresponds to a codeword _{Sm of} the m-th codebook; wherein, the codeword of the first-level codebook 4 The channel information can be independently characterized, indicating the index of the codeword in the codebook, and the m+1th codebook 4„ ₊ ι is generated according to the codeword 4„ of the mth codebook, m = \..M, M >\

The codeword 4 of the Mth codebook indicates a channel information;

The step of the transmitting end acquiring the channel information according to the channel indication information fed back by the receiving end comprises: the sending end acquiring the channel information according to the codeword A _{MSM of} the Mth-level codebook.

The method may further have the following feature, the m + 1 level codebook ₊₁ = ^, 0^), the F is a mapping relationship or a function relationship, and Q ^ is a codebook.

The method may further have the following feature: when the F is a mapping relationship, the C _bm is a solid Fixed codebook, alignment of multiple codeword information.

The method may further have the following feature, when the F is a function relationship,

It is a codebook composed of code words distributed around a center.

The method may further have the following feature: after receiving the channel indication information of the M codewords 4^... A _M , _SM , the sending end acquires the channel information according to the m-th channel indication information The codeword d obtains the codebook according to the codebook according to the first-order channel indication information, and acquires the codebook 4 according to the codebook 4 according to the first-order channel indication information;

After 4 ^ obtained codebook, according to feedback from the receiver of Μ stage channel indication information acquiring a codeword from the codebook ^ _¾ in.

The method may further have the following feature: the first level codebook is a codebook specified in a long term evolution system standard protocol; or, the codebook specified by the long distance evolution system standard protocol is subjected to a left multiplication matrix , ie 4 = ^D , D is the codebook specified in the Long Term Evolution System Standard Protocol, and R is the channel correlation matrix.

The invention also provides a system for multi-level feedback of channel information, comprising:

a receiving end, configured to feed back at least two levels of channel indication information to the transmitting end;

The sending end is configured to acquire channel information according to the channel indication information fed back by the receiving end. The system may further have the following feature: the receiving end is configured to: feed back M-level channel indication information to the transmitting end, where the M>1, the m-th channel indication information corresponds to a codeword of the m-th level codebook m = \..M , indicating an index of the codeword in the codebook; wherein, the codeword of the codebook 4 of the first level can independently represent the channel information; M codewords of the M-level codebook..... ^ _M} indicates a common channel information is _{_{Ω (4 ¾ -, ¾.}} ), as a function;

It is set to the sending end, according to the channel information acquiring _{Ω (4, ..... 4 ^ ¾} ).

The system may further have the following feature: the receiving end is configured to: feed the level channel indication information to the sending end, where the Μ > 1 , the channel indicating information of the mth level corresponds to a code of the mth level codebook 4 The word Λ^; wherein, the codeword of the first level codebook 4 can independently represent the channel information, indicating the index of the codebook 4 „ middle codeword, and the m+1th codebook according to the code of the mth codebook 4 4 _a word generated, m = \ · .Μ, Μ >1; [mu] of the first stage codebook codeword 4 _¾ information indicating a channel; The sending end is configured to: according to the codeword 4 _{s of} the Mth-level codebook, obtain channel information and feed back through the basic codebook and the incremental codebook combination, thereby facilitating more flexible feedback at different levels. Improve the overall quantization accuracy while controlling the feedback overhead; and the advanced codebook is based on the previous codebook.

Computational complexity of words preferred embodiment of the invention

The present invention provides a method for multi-level feedback of channel information, including:

The receiving end feeds back the M-level channel indication information to the sending end;

The transmitting end acquires channel information according to the M-level channel indication information.

Specifically divided into two ways to achieve:

(i) The first case:

The feedback from the receiver M-ary channel indicating information to the transmitting side, the M> 1, the m-th channel indicating information corresponding to the m-level codebook of one codeword ^, _m = l ..M, S m represents a code a _m in this codeword index; wherein the first-stage codebook codeword 4 independent information channel can be characterized; in the m-level codebook of m codewords ..... ^ _m} indicates a common channel information Ω(Α _]Λ ..... A _MS , Ω is a function, and the transmitting end acquires channel information according to Ω(4, ..... U. (Note: The first case)

The Ω is left multiplication, right multiplication, dot multiplication, difference, and or a combination thereof. The m + 1 level codebook 4 ^ is independent of the code word of the mth level codebook 4^ , the m + 1 level codebook

Λ ₊₁ has a center ₊₁ , the ₊₁ is an arbitrary matrix or vector, and the 4^ includes or does not include the center ₊₁ .

(ii) The second case:

The receiving end feeds the M-level channel indication information to the transmitting end, where the M>1, the m-th channel indication information corresponds to a codeword of the m-th codebook; wherein, the codeword of the first-level codebook 4 can Independently characterizing the channel information, indicating the index of the codeword in the codebook, the m+1th codebook 4„ ₊ ι is generated according to the codeword of the mth codebook A, m = lM, M> ]/, The codeword 4 _M 3⁄4 of the _Mth codebook indicates a channel information; the transmitting end acquires channel information according to the codeword A _{MSM of} the Mth codebook.

The m + 1 level codebook ^F (A _MSM , C _BM ), the F is a mapping relationship or a function relationship, and C _3⁄4m is a codebook.

The F is a mapping relationship, the ^^ is a fixed codebook, 4 _A C _bm corresponding to a plurality of codeword information.

When F is a functional relationship, the 4 ₊₁ =^)^„ is a codebook composed of codewords distributed around a center.

After acquiring the channel indication information of the M codewords 4 _Sl ..... A _M ^ _SM , the transmitting end acquires the codeword 4^ from the codebook according to the channel indication information of the mth level, according to the codebook ^Get the codebook Λ ₊₁ , initially, according to the first-level channel indication information, obtain the codebook according to 4 and the codebook 4 from the codebook 4;

After obtaining the codebook 4, the codeword is obtained from the codebook 4 according to the M-level channel indication information fed back by the receiving end.

In the first case and the second case, the first level codebook 4 is a codebook specified in a long term evolution system standard protocol; or, the codebook specified in the long term evolution system standard protocol is subjected to a left multiplication matrix. The latter codebook, ie 4=^D, D is the codebook specified in the Long Term Evolution System standard protocol, and R is the channel correlation matrix.

The present invention will be further described below by taking secondary feedback as an example.

The feedback of the channel characteristic information includes the quantization information capable of independently characterizing the feature vector, indicating the precoding indication (PMI) information of a certain codeword 4 in the primary codebook A and another codeword in the second secondary codebook B. Precoding Indication (PMI) information.

The first level codebook ^4 may be any codebook, and the A codebook may further be a codebook specified in the LTE standard protocol, and 4 may further be any codebook specified in the LTE standard protocol 36.211. The codebook may further be a codebook A = R after a certain codebook D passes through the left multiplying matrix in the prior art 2 (ie, the technique of rotating the LTE codebook by using the correlation matrix information of the channel). D can also be the codebook specified in the LTE standard protocol.

The first meaning of the feedback is: The first and second feedbacks collectively indicate that one feature vector quantization information is Ω(4, Α), where Ω represents a function. The meaning of the feedback is determined jointly by 4 and . to obtain the final eigenvector quantized representation. Different 4 identical indications of different codewords, the same 4 different indications of different codewords. Ω can represent a variety of functional relationships, such as left multiplication, right multiplication, point multiplication, and difference, or sum, and the combination of various functional relationships, not limited to this.

The function Ω can be: According to 4 and a function Ψ get a matrix Ψ (4), and then use Ψ (4) to perform left multiplication, right multiplication or dot multiplication operations. The matrix ψ(4) may further be a unitary matrix. The 酉 matrix may further be that the front r column is 4 (r is the Rank number);

Alternatively, the function Ω can also be:

According to S and a function, a matrix is obtained, and then 4 is subjected to left multiplication, right multiplication or dot multiplication. The matrix may further be a unitary matrix.

It is also possible to perform a left multiplication, a right multiplication or a dot multiplication operation directly on the 4th.

Wherein, the codebook B can exist independently of 4, and there is a clearly indicated center o. When the codebook B does not have a clearly indicated center, it can also find a vector or matrix 0 of the same dimension, such that the distance of the codeword to 0 is the same or 0. Codebook B can contain 0 or not. 0 can be any vector or matrix, and further can have only one matrix or vector of non-zero elements for each column.

The second case of feedback meaning representation:

The feature vector quantization information is characterized as A, wherein some or all of the codewords in codebook B can be expressed as B = F( ,, C _b ,

F may be a mapping relationship; through 4, and the mapping relationship F, can be found in some or all of the codeword from the codebook is defined as C _¾ part B or all of the code words. When F is expressed as a mapping relationship, C _{6 is} understood to be a fixed codebook. Each 4 corresponds to a plurality of medium codeword information. Further, according to the information RI ( Rank Indicator) of the Rank and combined with other feedback information, 4 may correspond to different sets of code words in C _h . That is to say, for different 4, the meaning of the second level feedback is different.

Alternatively, F can also be expressed as a functional relationship; according to 4 and codebook C ₆ and F can be determined A codebook: 6. The, F can be further operable by a left-5 = (4) C _¾, of the possible further to a unitary matrix, the unitary matrix may be further classified. 4 is a front r (r is

Rank number). (It is constructed in the center. The codebook consisting of evenly distributed codewords around.

The first-level codebook can be regarded as a basic codebook, and can be independently configured as a feedback mode for feature vector quantization, but the quantization precision is low. However, it can be better suited to SU-MIMO, the second-level codebook, the third-level codebook, and the higher-level codebook can be regarded as an incremental codebook or an accuracy adjustment codebook.

In the second case, compared with the first case, the codeword feedback in the codebook B obtained according to 4 can be directly expressed as feature vector quantization information, and in the first case, it must be combined with 4 to represent the accurate feature vector. Quantify information.

Similarly, feedback from Level 3 or higher codebooks has two corresponding cases.

The first case:

The information represented by the level 3 feedback is a code word in the level 3 codebook B. The meaning of the feedback representation is that the first, second and third levels of feedback together indicate that one feature vector quantization information is r(4, , ς), where r represents a function. The meaning of the feedback is based on 4, . and jointly determined to obtain the final eigenvector quantized representation. Different identical c _p indicate different code words.

r can represent a variety of functional relationships, such as left multiplication, right multiplication, point multiplication, and difference, or a combination of sum and operation, not limited to this.

The second case:

Some or all of the codewords in the second codebook B may be expressed as B^^' Cj, and some or all of the codewords in the third-level codebook C may be expressed as C = ( , C ₆ )

F is a mapping relationship or a function, and G may be the same as or different from C _fc .

The feedback frequency of the first 1, 2, 3 or higher feedback may be configured and may be different. A typical example is that the base codebook (primary codebook) has a slow update frequency, for example, 100ms update, and the level 2 and level 3 codebooks can be updated faster because they belong to the codebook fine-tuning, for example, 5ms. Update once. Of course, the practice of updating at the same time is not excluded.

According to the different configuration feedback rates of the hierarchical codebook, there are several basic modes. The first one only has the basic codebook feedback, the second has the basic codebook feedback and the incremental codebook feedback, the basic codebook feedback and the incremental codebook feedback. Forms a complete codebook feedback, and this complete codebook feedback can be cleared Clearly split into basic codebook feedback and incremental codebook feedback, the base codebook can be used separately, and prior art scheme 1 (ie, independently designing a set of high-overhead codes according to the required MU_MIMO codebook overhead) In this scenario, it is not possible to split a base codebook that can be used separately from a complete codebook.

The feedback mode of the present invention can strongly support the SU-MIMO and MU-MIMO dynamic switching transmission modes.

The following describes the specific implementation method by taking the Rank1 and Rank2 channel information feedback as an example, but the present invention is not limited to these two cases.

Example 1

In this embodiment, the multi-level channel information feedback method provided by the present invention includes:

Step 110: Determine a Rank of the channel, and select a corresponding codeword from the codebook corresponding to the Rank value.

4, as the first level feedback.

The first-level codebook uses the LTE codebook in Table 2, or the first-level codebook uses the LTE codebook rotated by the channel correlation matrix in the prior art, and the quantization and feedback method is in the SU-MIMO mode in LTE. The feedback method is the same.

Wherein, according to the feature vector information Eigenvector, a suitable codeword 4 can be selected as the first feature vector from the first-level codebook A according to the principle of maximum chord distance, maximum throughput, or maximum signal-to-interference ratio (SINR). Level quantification.

In step 120, the second level feedback is the PMI information of the codeword in another codebook B. After the UE selects and receives the form feedback through PMI, the method selected by the UE is:

a) Fixed or dynamically generate a secondary codebook B, and define a codeword distribution center 0, 0 may or may not be included in B. Usually 6> can be selected as a matrix or vector with only one non-zero element in each column. The typical method of 0 is:

b) Determine a matrix Rot by 4 and 0, which can be rotated to Ο after multiplying A _t by Rot, and the same operation can be performed for the feature vector or the matrix formed by it Eigen Vector = Rotx Eigen Vector

Since 4 and Eigenvector are multiplied by the same matrix, their error properties are not changed. So Eigenvector is rotated around 0. By codewords distributed around O, a codeword can be found, and the O error relationship "is the most similar or identical to the error relationship of ^wi^ctor and 4".

Step 130: After the base station obtains 4 information and information, the inverse operation T ¹ of the rotation in step 120 is performed, and T ^{1 is} determined by the first level feedback information 4 and o. The feature vector information with higher quantization precision can be obtained by the inverse operation.

Example 2

Step 210: Determine a Rank of the channel, and select a corresponding codeword 4 from the codebook corresponding to the Rank value. As level 1 feedback.

The first level quantization 4 of the feature vector can be found from the first level codebook A based on the feature vector information _g i3⁄4ctor of the channel matrix.

The quantization method and the feedback method are the same as the feedback method in SU-MIMO mode in LTE except that the codebook is different.

The first-level codebook uses the LTE codebook in Table 2, or the first-level codebook uses the LTE codebook rotated by the channel correlation matrix in the prior art, that is, the codebook using the left-left LTE is used to obtain a new codebook. , R is the channel correlation matrix.

Step 220: The second level feedback is PMI information of the codeword in another codebook B. After the UE selects and receives the form feedback through PMI, the definition of codebook B and the method for UE selection are as follows: A codeword is found from the second-level codebook B, which may be a fixed rotation matrix of N*N, or may be passed through a A rotation matrix obtained by the function, for example:

K{AM = te, (3⁄4 denotes a multi-column vector orthogonal to 4.

The codeword found by the UE is the codeword closest to the feature vector g cto after the left multiplication rotation

Step 230: After the base station learns the 4 information, it can obtain the same operation by using the same operation. After the B ₃ is rotated by 4, the feature vector information with higher quantization precision can be obtained.

Example 3 Step 310: Determine a Rank of the channel, and select a corresponding codeword A _t from the codebook corresponding to the Rank as the first level feedback.

For example, the first-order quantization 4 of the feature vector is found from the first-order codebook A based on the feature vector information 3⁄4ctor.

The first-level codebook uses the LTE codebook in Table 2, or the first-level codebook uses the LTE codebook rotated by the channel correlation matrix in the prior art, the quantization and feedback method, and the feedback method in the SU-MIMO mode in LTE. the same.

Step 320, according to the information of 4, can find some codebooks B composed of codewords distributed around 4, and then find the feature vector g ecto from B to more accurately represent the codeword.

The method is constructed by a baseline codebook 0_ and a 4 joint. For example, construct some centers. Codewords evenly distributed around constitute one codebook c _6. According to 4 and. The relationship determines a matrix R, and R is multiplied by Q_ to obtain a codebook consisting of codewords distributed around 4 in the sentence.

B. Select a codeword with the smallest quantization error from Eigenvector from B as the quantized information of Eigenvector.

Ratio 3⁄4P, i? = HH([l 0 0 0j -U{A _t )) , HH shows Household, which means that the same phase adjustment operation is performed on each element of 4 such that its first element is a real number. This way you can get an R matrix, so that. Selected to 4.

Use the same ^? to perform the same left multiplication operation for each codeword in 0_ to obtain a surrounding codebook B, select a codeword B from B as the quantization information of Eigenvector, and feed back the corresponding PMI.

Step 330: The base station according to the information of 4 and the specified. And the same R matrix determination method, you can get the codebook B around 4, and find the codeword in it as the final representation of the feature vector.

Example 4

Step 410: Determine a Rank of the channel, and select a corresponding codeword A from the codebook corresponding to the Rank. As level 1 feedback.

The UE finds the first level quantization 4 of the feature vector from the first level codebook A according to the feature vector information Eigenvector. The first-level codebook uses the LTE codebook in Table 2, or the first-level codebook uses the LTE codebook rotated by the channel correlation matrix in the prior art, the quantization and feedback method, and the feedback method in the SU-MIMO mode in LTE. the same.

Step 420, using the mapping relationship, find (: some codewords corresponding to 4 constitute codebook B, and determine the numbering rule. Find a codeword from B as a representation of the feature vector. And use PMI to perform 5 _; Feedback.

Wherein, some codewords corresponding to 4 in the codebook c are found according to 4 and the mapping relationship, and form part of the codebook B or B. B contains N code words that are very close to the 4 strings. For example, N can be a power value of 2, 8, 16, etc.

Step 430: After obtaining the first level feedback 4, the base station may use the same mapping relationship to find the codebook.

B, and according to the same numbering rule and the PMI information fed back by the UE, find a representation as a feature vector.

The sending end is configured to acquire channel information according to the channel indication information fed back by the receiving end. For example, the receiving end feedback method and the method for the sending end to obtain the channel information are described in the method embodiment, and details are not described herein again.

The present invention can not only solve the problem of feedback accuracy from single-user MIMO to multi-user MIMO, but can also be used to improve the feedback accuracy from single-user MIMO to multi-user MIMO.

The invention provides a more flexible feedback of different levels through the combination of the base codebook and the incremental codebook, which can effectively control the feedback overhead and ensure the maximum backward compatibility while improving the overall quantization precision. The advanced codebook can significantly reduce the computational complexity of the codeword based on the design of the previous codebook information.

One of ordinary skill in the art will appreciate that all or a portion of the above steps may be performed by a program to instruct the associated hardware, such as a read only memory, a magnetic disk, or an optical disk. Alternatively, all or part of the steps of the above embodiments may also be implemented using one or more integrated circuits. Correspondingly, each module/unit in the foregoing embodiment may be implemented in the form of hardware, or may be implemented in the form of a software function module. The invention is not limited to any specific form of combination of hardware and software. Industrial applicability

The method for multi-level feedback of channel information provided by the invention has more flexible feedback of different levels through the combination of the basic codebook and the incremental codebook, and controls the feedback overhead while improving the overall quantization precision; The method based on the design of the previous level codebook information significantly reduces the computational complexity of the codeword compared to the independent codebook design in the prior art.

Claims

Claim

A method for multi-level feedback of channel information, comprising:

The receiving end feeds back at least two levels of channel indication information to the transmitting end;

The transmitting end acquires channel information according to the channel indication information fed back by the receiving end.

2. The method according to claim 1, wherein the step of the receiving end feeding back at least two levels of channel indication information to the transmitting end comprises:

The feedback from the receiver M-ary channel indicating information to the transmitting side, the M> 1, the m-th channel indicating information corresponding to the m-th stage codebook codeword A _m A A _{M Sm,} m = l..M, S _m represents an index of a codeword in the codebook A _m ; wherein, the codeword of the first-level codebook 4 can independently represent the channel information;

The M code words of the M-level codebook ..... ^ _M } collectively indicate that one channel information is

Ω ( ..... A _{M SM} ) , Ω is a function, the step of the transmitting end acquiring the channel information according to the channel indication information fed back by the receiving end comprises: the transmitting end according to Ω(4,, .. ... Λ^ _Μ ) Get channel information.

3. The method of claim 2, wherein the Ω is one of the following operations or a combination thereof: left multiplication, right multiplication, dot multiplication, difference, and sum operation.

4. The method according to claim 2 or 3, wherein the m + 1 level codebook 4 „ _{+1 is} independent of the code word d of the mth codebook

The method according to claim 4, wherein the m + 1 level codebook ₊₁ exists in a center, and the matrix is an arbitrary matrix or vector, and the center or the _{+1 is} included Except the center. A code word other than _m+1 and the center. The distance of _m+1 is the same.

6. The method according to claim 1, wherein the step of the receiving end feeding back at least two levels of channel indication information to the transmitting end comprises:

The receiving end feeds the M-level channel indication information to the transmitting end, where the M>1, the m-th channel indication information corresponds to a codeword of the m-th codebook, where the codebook of the first-level codebook 4 The codeword can independently represent the channel information, and represents the index of the codebook 4 „中码字, the m ₊ 1th codebook 4„ ₊ ι is generated according to the codeword of the mth codebook, m = \..M, M >\;

The codeword _{4⁄4 of} the Mth codebook indicates a channel information;

Step step of obtaining, by the transmitting end, channel information according to the channel indication information fed back by the receiving end The transmitting end acquires channel information according to the codeword A _{MSM of} the Mth-level codebook.

7. The method of claim 6, wherein the m + 1 level codebook

A _M+ , =F{A _MSM , C _M ), the F is a mapping relationship or a functional relationship, which is a codebook.

8. The method according to claim 7, wherein, when F is a mapping relationship, the ς^ is a fixed codebook, and 4^ corresponds to a plurality of codeword information in _m .

9. The method according to claim 7, wherein, when the F is a functional relationship, the

_Ra is a codebook composed of code words distributed around a center.

10. The method of claim 6, the method further comprising:

After acquiring the channel indication information of the M codewords 4..., the transmitting end acquires the codeword from the codebook according to the m-th channel indication information, and obtains the codebook Λ ₊₁ according to the codebook ,, Initially, the codebook 4 is obtained from the codebook 4 according to the level 1 channel indication information according to 4^; and the codebook 4;

After obtaining the codebook, the codeword _3b is obtained from the codebook according to the feedback information of the first-level channel fed back by the receiving end.

The method according to claim 2 or 6, wherein the first level codebook 4 is a codebook specified in a long term evolution system standard protocol; or, the codebook specified in the long term evolution system standard protocol passes through the left The codebook after multiplying the matrix, ie 4=^D, D is the codebook specified in the long-term evolution system standard protocol, and R is the channel correlation matrix.

12. A system for multi-level feedback of channel information, comprising:

The sending end is configured to acquire channel information according to the channel indication information fed back by the receiving end.

13. The system of claim 12, wherein

The receiving end is configured to feed back M-level channel indication information to the transmitting end, where the M>1, the m-th channel indication information corresponds to a codeword of the m-th codebook, 4 _A , m = \..M , indicating an index of a codeword in the codebook; wherein, the codeword of the codebook of the first level codebook 4 can independently represent the channel information; the M codewords {.....} of the M-level codebook collectively indicate a channel information For A-. υ, is a function; It is set to the sending end, according to the channel information acquiring _{Ω (4, ..... 4 ^ ¾} ).

14. The system of claim 12, wherein

The receiving end is configured to: feed the level channel indication information to the sending end, where the Μ>1, the channel indicating information of the mth level corresponds to one codeword of the mth codebook; wherein, the first level codebook 4 The codeword can independently represent the channel information, and represent the index of the codeword in the codebook, and the m+1th codebook is generated according to the codeword ^ of the mth codebook, w = l..iJ, M>l; The codeword of the Mth codebook, _3⁄4 indicating a channel information;

The transmitting end is configured to acquire channel information according to the codeword A _{M SM of} the Mth-level codebook.