CN103368628B - Double-current beam forming method based on code books in TD-LTE system - Google Patents

Abstract

Provided is a double-current beam forming method based on code books in a TD-LTE system. The double-current beam forming method includes the steps of selecting TD-LTE 8 antenna code books, carrying out antenna correction compensation, computing a channel estimated value H after the antenna correction compensation according to an original channel estimated value and an up-going correction weight, judging whether user terminal antennas are transmitted in turns, if yes, judging whether the H is successfully matched, if yes, combining a matrix [H1,H2] and estimating normalized initial phases of M RB channels according to antennas, and changing the combined matrix into [H(1), H(2)], judging granularity of a figurative weight, setting the weight granularity to be m RB, if m＞1, averaging channel estimated values of m RB and searching the double-current beam forming code book weight; if m=1, searching the double-current beam forming code book weight, traversing all RB and then finishing operation. According to the double-current beam forming method based on the code books in the TD-LTE system, the beam forming code book is preset on a base station side, two orthorhombic code book weights are searched according to user channel information, therefore, quick forming is achieved, and wireless performance of the TD-LTE system is improved.

Description

Codebook-based double-current beam forming method in TD-LTE system

Technical Field

The invention belongs to the technical field of antennas, and particularly relates to a codebook-based double-current beam forming method in a TD-LTE system.

Background

In 2000, the TD-SCDMA mobile communication standard which is the leading standard in China is officially accepted by ITU and 3GPP as one of the 3G international standards, which is a major breakthrough in the technical history of telecommunication in China, the TD-LTE standard which is a subsequent evolution scheme of the TD-SCDMA standard is also favored by the mobile communication industry, and under the positive promotion of the government and related manufacturers in China, the TD-LTE-Advanced submitted in China is officially listed as the IMT-Advanced (4G) international standard through evaluation. Compared with the 3G technology, the LTE system has the following characteristics: (1) the communication rate and the spectrum efficiency are improved; the method comprises the following steps of (1) ensuring the service quality, expanding the service providing capability, providing more services with lower cost and better user experience, (3) more flexibly configuring the system bandwidth, (4) definitely proposing that the system provides an optimized condition for low-speed users and simultaneously improves the throughput of cell edge users on the basis of supporting high-speed movement.

The mimo technology is a technology for performing spatial diversity using multiple transmit and receive antennas. It uses discrete multi-antenna, which can effectively decompose the communication link into many parallel sub-channels, thus greatly increasing the capacity. In the downlink of TD-LTE, the multi-antenna transmission mode mainly includes technologies such as transmit diversity, beamforming, space-time and coding, and multi-user multiple-input multiple-output. The TD-LTE system supports rich application scenarios, and defines 9 transmission modes of downlink services to ensure that a technical scheme for performance optimization can be selected in various environments. Wherein, the transmission mode 7, the transmission mode 8 and the transmission mode 9 are beam forming modes.

The working principle of the beamforming is to generate a radiation pattern with a certain directivity by utilizing the strong correlation of a spatial channel and the interference principle of waves, so that the main lobe of the radiation pattern is self-adaptively pointed to a user, thereby improving the signal-to-noise ratio and obtaining obvious array gain. The purpose of beamforming is to expand the signal coverage, improve the edge throughput, and suppress interference. How to effectively and efficiently generate the beam forming weight to match the change of the channel so as to improve the throughput of the system, and the method becomes the unique standard for measuring the beam forming algorithm.

The LTE R9 extends beamforming to dual-stream transmission, thereby realizing the combination of beamforming and spatial multiplexing techniques. The double-current beam forming technology is applied to the condition that signal scattering bodies are sufficient, the intelligent antenna technology and the multi-input multi-output spatial multiplexing are combined, the characteristics of wide coverage, large cell capacity and small interference of the traditional single-current beam forming technology can be kept, the reliability of edge users can be improved, and meanwhile the throughput of central users of the cell can be effectively improved. A new dual-port dedicated pilot is defined in LTE R9 and new control signaling is introduced, with the antenna configuration being 8 x 2.

The calculation complexity is higher based on the method that singular value decomposition is mainly adopted for downlink beamforming of the TD-LTE system at present.

Disclosure of Invention

The technical problem to be solved by the invention is to overcome the defects of the prior art and provide a codebook-based double-current beamforming method in a TD-LTE system, which is simple, can realize double-current beamforming, has low complexity of double-current beamforming and low information feedback between subsystems.

The technical scheme adopted for solving the technical problems comprises the following steps:

on the basis of single-stream beamforming based on a codebook, a double-stream beamforming method based on the codebook is provided, the method reduces the complexity of double-stream beamforming of a TD-LTE system relative to a common singular value decomposition method, and a beamforming codebook of a TD-LTE8 antenna is designed as follows:

wherein,

k is a natural number of 0 to 15,

v_i＝[1 e^j2π*2*i/32e^j4π*2*i/32e^j6π*2*i/32]^T

v_j＝[1 e^j2π*2*j/32e^j4π*2*j/32e^j6π*2*j/32]^T

wherein i is 0,1, …, 15; j is 0,1, …, 15.

The TD-LTE8 antenna beamforming codebook selects a total of 16 × 16 × 16.

1. Antenna correction and compensation for 8 antenna of TD-LTE systemBased on the original channel estimateCalculating the channel estimation after antenna correction compensation according to the uplink correction weightWhere n is the RB number index, ka_RxThe index is an antenna index, u is a user index, i is a transmitting antenna index of the user u, and the value of i is 1 or 2; RB is a resource block;

2. and (3) judging whether the user terminal antenna is transmitted in turn, if so, turning to the step (3), and if not, turning to a single-flow weight to construct a double-flow weight state.

3. And judging whether the user channel matrix H is successfully paired, wherein H is an 8 multiplied by 1 channel matrix, if the user channel matrix H is successfully paired, turning to the step 4, and if the user channel matrix H is not successfully paired, turning to a single-stream weight to construct a double-stream weight state.

4. Combined matrix [ H₁,H₂]And normalizing the initial phase for the channel estimation values of M RBs according to the antennas, respectively, and the combination matrix becomes [ H ]₍₁₎,H₍₂₎]；

5. Judging the granularity of the shaped weight, setting the granularity of the shaped weight as m RBs, averaging the channel estimation values of the m RBs if m is greater than 1, and searching the weight of the double-current beam-forming codebook; and if m is 1, directly searching the weight of the double-current beamforming codebook.

6. And ending after traversing all RBs.

The method has the advantages of simply and efficiently realizing the double-current beam forming of the TD-LTE system, reducing the information feedback among TD-LTE subsystems and reducing the complexity of system realization.

Drawings

Fig. 1 is a flowchart of overall processing of dual-stream beamforming based on a codebook in a TD-LTE system.

Fig. 2 is a diagram of the antenna configuration of the base station and the user terminal.

Detailed Description

The present invention will be described in further detail with reference to the drawings and examples, but the present invention is not limited to these examples.

Example 1

The method for double-current beamforming based on the codebook in the TD-LTE system of the embodiment includes the following steps:

the antenna arrangement of the base station and the user terminal is as shown in fig. 2.

This embodiment processes RB number M to take a value of 8.

The beam forming design of the TD-LTE8 antenna codebook is as follows:

k is a natural number of 0 to 15,

v_i＝[1 e^j2π*2*i/32e^j4π*2*i/32e^j6π*2*i/32]^T

v_j＝[1 e^j2π*2*j/32e^j4π*2*j/32e^j6π*2*j/32]^T

wherein i is 0,1, …, 15; j is 0,1, …, 15;

the codebook selects a total of 16 × 16 × 16.

The first step is as follows: start 11, antenna calibration Compensation 12, for TPerforming antenna correction compensation on 8 antennas of the D-LTE system according to the original channel estimation valueCalculating the channel estimation after antenna correction compensation according to the uplink correction weight

Wherein n is the RB number index ka_RxThe index is an antenna index, u is a user index, i is a transmitting antenna index of the user u, and the value of i is 1 or 2; RB is a resource block; (the invention adopts the processing mode consistent with the single antenna transmission at the antenna side of the user terminal)

Note: uplink correction weightThe calculation method comprises the following steps that after the average H of each subcarrier on each channel is obtained, the amplitude-phase characteristics of each subcarrier are obtained, and by taking a certain channel as a reference, the amplitude-phase inconsistency of each channel relative to the reference channel is calculated by an online correction method, so that the compensation coefficient, namely the correction weight, of each antenna channel is obtained;

the second step is that: judging whether the user terminal antenna is sent out in turn 13, if the user terminal antenna is sent out in turn, switching to the third step, and if the user terminal antenna is not sent out in turn, switching to a single-flow weight construction double-flow weight state 14;

the third step: judging whether H is successfully paired 15, if so, switching to the fourth step, and if not, switching to a single-flow weight construction double-flow weight state 16; the method for constructing the double-flow weight by the single-flow weight is as follows;

setting the weight granularity m as 1RB or setting the weight granularity m as 1RB

1. Weighted by a single streamSearch method acquisition, PMI₁(I₁,J₁,K₁)；

2. Querying and searching a predetermined codebook, finding and PMI₁(I₁,J₁,K₁) Orthogonal elements, constituting a codebook set Q, assuming that there are L elements in this codebook set;

3. performing loop traversal search L times on L codebooks in a codebook set Q, and searching for i, j, k which satisfy the following formula and are recorded as PMI₂(I₂,J₂,K₂)

Wherein, when m is 1, the cyclic Loop is 1: m, when M >1, Loop ═ 1: [ M/M ], in the embodiment, the value of M is 8, and other positive integers can be adopted;

the fourth step: combined matrix [ H1, H2]17, normalizing the channel estimation values of M RBs according to the antennas respectively to obtain an initial phase, and changing the combination matrix into H₍₁₎,H₍₂₎]18, the specific method is as follows:

1. determine the phase θ of the first antenna per RB

2. Per RB normalization by antenna 1 phase

Wherein, the cyclic Loop is 1: m;

the fifth step: judging the granularity 19 of the shaped weight, setting the granularity of the shaped weight as m RBs, if m is greater than 1, firstly averaging the channel estimation values of the m RBs (step 20), and then searching a double-current beam-shaped codebook weight 21; if m is 1, directly searching a double-current beamforming codebook weight 24, wherein the specific method is as follows;

when the weight granularity m is 1RB

1. In order to simplify the search space, the search is performed in 3 groups of 16 codebooks of i, j and k, and PMI serial numbers i, j and k satisfying the following formula are respectively searched and recorded as PMIs₁(I₁,J₁,K₁)25

Search 16 times to obtain I₁Wherein index1 is 4 antennas corresponding to port 0;

search 16 times to obtain J₁；

Search 16 times to obtain K₁；

Wherein index2 is 4 antennas corresponding to port1, so far, PMI₁(I₁,J₁,K₁) Has already searched for;

2. querying and searching a predetermined codebook, finding and PMI₁(I₁,J₁,K₁) Orthogonal elements form a codebook set Q, and L elements are assumed in the codebook set Q;

3. circulating L codebooks in a codebook set Q, traversing and searching for L times, searching PMI serial numbers i, j and k meeting the following formula and marking as PMI₂(I₂,J₂,K₂)

When the weight granularity m >1RB

1. Channel estimation H for m RBs_(i)The values are averaged and the calculation formula is as follows

Wherein jj is 1, 2, …, Ka_Tx，Ka_TxFor the base station transmitting antenna index, i is the user transmitting antenna index, i takes 1 or 2, and the cyclic Loop is 1: [ M/M ]]；

2. To simplify the search space, 3 sets of 16 codebooks of i, j, k are searched, and PMI numbers i, j, k 22 satisfying the following formula are searched respectively

Search 16 times to obtain I₁In, index1 is 4 antennas corresponding to port 0;

search 16 times to obtain J₁In, index2 is 4 antennas corresponding to port 1;

search 16 times to obtain K₁；

Wherein, the cyclic Loop is 1: [ M/M ]

3. Querying and searching a predetermined codebook, finding and PMI₁(I₁,J₁,K₁) Orthogonal elements form a codebook set Q, and L elements are assumed in the codebook set Q;

4. circulating L codebooks in a codebook set Q, traversing and searching for L times, searching for i, j and k meeting the following formula, and recording the i, j and k as PMI₂(I₂,J₂,K₂)

Wherein, the cyclic Loop is 1: [ M/M ]

And a sixth step: and judging whether all RBs 23, 26 are traversed or not, and finally ending 27.

Claims (1)

1. A double-current beam forming method based on a codebook in a TD-LTE system is characterized by comprising the following steps:

selecting a TD-LTE8 antenna beamforming codebook:

k is a natural number of 0 to 15,

v_i＝[1 e^j2π*2*i/32e^j4π*2*i/32e^j6π*2*i/32]^T

v_j＝[1 e^j2π*2*j/32e^j4π*2*j/32e^j6π*2*j/32]^T

wherein i is 0,1, …, 15; j is 0,1, …, 15;

the TD-LTE8 antenna beam forming codebook selects a total of 16 × 16 × 16;

(1) performing antenna correction compensation on 8 antennas of the TD-LTE system according to the original channel estimation valueCalculating the channel estimation after antenna correction compensation according to the uplink correction weightWhere n is the RB number index, ka_RxThe index is an antenna index, u is a user index, i is a transmitting antenna index of the user u, and the value of i is 1 or 2; RB is a resource block;

(2) judging whether the user terminal antenna is transmitted in turn, if so, turning to the step (3), and if not, turning to a single-flow weight to construct a double-flow weight state;

(3) judging whether the user channel matrix H is successfully paired, wherein H is an 8 multiplied by 1 channel matrix, if the user channel matrix H is successfully paired, the step (4) is carried out, and if the user channel matrix H is not successfully paired, the step is carried out, and the step is carried out, the step is;

(4) combined matrix [ H₁,H₂]And normalizing the initial phase for the channel estimation values of M RBs according to the antennas, respectively, and the combination matrix becomes [ H ]₍₁₎,H₍₂₎]；

The specific method comprises the following steps:

a. determine the phase θ of the first antenna per RB

$\mathrm{\θ}=arctan\left(\frac{imag\left(\stackrel{\‾}{H_i}\right(1,Loop))}{real\left(\stackrel{\‾}{H_i}\right(1,Loop))}\right)$

b. Per RB normalization by antenna 1 phase

$H_{\left(i\right)}({\mathrm{ka}}_{Rx},Loop)=\frac{\stackrel{\‾}{H_i}({\mathrm{ka}}_{Rx},Loop)}{\cos \mathrm{\θ}+j\sin \mathrm{\θ}}$

Wherein, the cyclic Loop is 1: m;

(5) judging the granularity of the shaped weight, setting the granularity of the shaped weight as m RBs, averaging the channel estimation values of the m RBs if m is greater than 1, and searching the weight of the double-current beam-forming codebook; if m is 1, directly searching a double-current beam forming codebook weight;

when the weight granularity m is 1 RB:

a. in order to simplify the search space, the search is performed in 3 groups of 16 codebooks of i, j and k, and PMI serial numbers i, j and k satisfying the following formula are respectively searched and recorded as PMIs₁(I₁,J₁,K₁)

$G11\left(i\right)=\underset{i}{\arg }\left(max\right\{\[|{v_i}^T\×H_{\left(1\right)}(index1,Loop){|}^2+|{v_i}^T\×H_{\left(2\right)}\left(index1,Loop\right){|}^2\]\left\}\right)$

Search 16 times to obtain I₁Wherein index1 is 4 antennas corresponding to port 0;

$G12\left(j\right)=\underset{j}{\arg }\left(max\right\{\[|{v_j}^T\×H_{\left(1\right)}(index2,Loop){|}^2+|{v_j}^T\×H_{\left(2\right)}\left(index2,Loop\right){|}^2\]\left\}\right)$

search 16 times to obtain J₁；

$G1\left(k\right)=\underset{k}{\arg }\left(\max \right\{\[|{\left[\begin{array}{c}{v}_I\\ v_J{\mathrm{\φ}}_k\end{array}\right]}^T\×H_{\left(1\right)}(\[index1;index2\],Loop){|}^2+|{\left[\begin{array}{c}{v}_I\\ v_J{\mathrm{\φ}}_k\end{array}\right]}^T\×H_{\left(2\right)}(\[index1;index2\],Loop){|}^2\]\})$

Search 16 times to obtain K₁；

Wherein index2 is 4 antennas corresponding to port1, so far, PMI₁(I₁,J₁,K₁) Has already searched for;

b. querying and searching a predetermined codebook, finding and PMI₁(I₁,J₁,K₁) Orthogonal elements, groupsA codebook set Q is formed, and L elements are assumed in the codebook set Q;

c. circulating L codebooks in a codebook set Q, traversing and searching for L times, searching PMI serial numbers i, j and k meeting the following formula and marking as PMI₂(I₂,J₂,K₂)

$G2=\underset{i,j,k}{\arg }\left(\max \right\{\[|{\left[\begin{array}{c}{v}_i\\ v_j{\mathrm{\φ}}_k\end{array}\right]}^T\×H_{\left(1\right)}(\[index1;index2\],Loop){|}^2+|{\left[\begin{array}{c}{v}_i\\ v_j{\mathrm{\φ}}_k\end{array}\right]}^T\×H_{\left(2\right)}\left(\[index1;index2\],Loop\right){|}^2\]\left\}\right)$

② when the weight granularity m is more than 1RB

a. Channel estimation H for m RBs_(i)The values are averaged and the calculation formula is as follows

${\stackrel{\‾}{\stackrel{\‾}{H}}}_{\left(i\right)}(jj,Loop)=\frac{\underset{ii=1}{\overset{m}{\Σ}}{H}_{\left(i\right)}(jj,ii)}{m}$

Wherein jj is 1, 2, …, Ka_Tx，Ka_TxFor the base station transmitting antenna index, i is the user transmitting antenna index, i takes 1 or 2, and the cyclic Loop is 1: [ M/M ]]；

b. In order to simplify the search space, 3 groups of 16 codebooks of i, j and k are searched, and PMI serial numbers i, j and k meeting the following formula are respectively searched;

$G11\left(i\right)=\underset{i}{\arg }\left(max\right\{\[|v_i^T\×{\stackrel{\‾}{\stackrel{\‾}{H}}}_{\left(1\right)}(index1,Loop){|}^2+|v_i^T\×{\stackrel{\‾}{\stackrel{\‾}{H}}}_{\left(2\right)}\left(index1,Loop\right){|}^2\]\left\}\right)$

search 16 times to obtain I₁In, index1 is 4 antennas corresponding to port 0;

$G12\left(j\right)=\underset{j}{\arg }\left(max\right\{\[|v_j^T\×{\stackrel{\‾}{\stackrel{\‾}{H}}}_{\left(1\right)}(index2,Loop){|}^2+|v_j^T\×{\stackrel{\‾}{\stackrel{\‾}{H}}}_{\left(2\right)}\left(index2,Loop\right){|}^2\]\left\}\right)$

search 16 times to obtain J₁In, index2 is 4 antennas corresponding to port 1;

$G1\left(k\right)=\underset{k}{\arg }\left(\max \right\{\[|{\left[\begin{array}{c}{v}_I\\ v_J{\mathrm{\φ}}_k\end{array}\right]}^T\×{\stackrel{\‾}{\stackrel{\‾}{H}}}_{\left(1\right)}(\[index1;index2\],Loop){|}^2+|{\left[\begin{array}{c}{v}_I\\ v_J{\mathrm{\φ}}_k\end{array}\right]}^T\×{\stackrel{\‾}{\stackrel{\‾}{H}}}_{\left(2\right)}(\[index1;index2\],Loop){|}^2\]\})$

search 16 times to obtain K₁；

Wherein, the cyclic Loop is 1: [ M/M ]

c. Querying and searching a predetermined codebook, finding and PMI₁(I₁,J₁,K₁) Orthogonal elements form a codebook set Q, and L elements are assumed in the codebook set Q;

d. circulating L codebooks in a codebook set Q, traversing and searching for L times, searching for i, j and k meeting the following formula, and recording the i, j and k as PMI₂(I₂,J₂,K₂)

$G2=\underset{i,j,k}{\arg }\left(\max \right\{\[|{\left[\begin{array}{c}{v}_i\\ v_j{\mathrm{\φ}}_k\end{array}\right]}^T\×{\stackrel{\‾}{\stackrel{\‾}{H}}}_{\left(1\right)}(\[index1;index2\],Loop){|}^2+|{\left[\begin{array}{c}{v}_i\\ v_j{\mathrm{\φ}}_k\end{array}\right]}^T\×{\stackrel{\‾}{\stackrel{\‾}{H}}}_{\left(2\right)}\left(\[index1;index2\],Loop\right){|}^2\]\left\}\right)$

Wherein, the cyclic Loop is 1: [ M/M ];

(6) and ending after traversing all RBs.