CN112702093A - Channel estimation method in FDD downlink multi-user large-scale MIMO system - Google Patents

Abstract

The present invention provides a channel estimation method in an FDD downlink multi-user massive MIMO system, which includes steps: S1. Designing training sequences for multiple users and processing received signals; S2. Performing corresponding angle estimation according to the received signals; S3. Using angle estimation value to construct a channel steering vector; S4. Reconstruct downlink channel state information according to the constructed channel steering vector. The present invention is based on the channel estimation method of a small number of uniform training sequences, multiple users can estimate the channel state information at the same time, effectively reduce the overhead of pilot frequency, and lay a foundation for realizing higher system throughput.

Description

Channel Estimation Method in FDD Downlink Multi-User Massive MIMO System

technical field

The present invention relates to the technical field of wireless communication, in particular to a channel estimation method in an FDD downlink multi-user massive MIMO system.

Background technique

In a massive multiple-input multiple-output (MIMO) system, using the massive antenna array of the base station (BS), the system can obtain extremely high spatial resolution and spatial division multiplexing gain, and also can achieve high spatial resolution without severe interference. It can provide services for multiple users at the same time, which greatly improves the spectral efficiency and transmission rate of the system. However, these excellent performances all depend on the availability of channel state information (CSI), so how to obtain a complete CSI in an actual system is very important to the system performance.

For a frequency division duplex (FDD) massive MIMO system, a large amount of pilot overhead is required to obtain complete downlink CSI. This is because in practical massive MIMO systems, the training times and feedback overhead are proportional to the number of BS antennas, and traditional linear channel estimation methods such as least squares algorithm (LS) and linear least mean square error (LMMSE) are used. Algorithm) To obtain complete CSI resources, it is necessary to send training sequences with the same number of base station antennas, resulting in FDD downlink channel estimation methods developed in recent years, almost all of which are designed for different users’ channel information. Different training sequences to obtain CSI, so that , for a multi-user system, it will consume more time and resources, which is impractical.

In view of the above technical problems, it is necessary to improve it.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a channel estimation method in an FDD downlink multi-user massive MIMO system in view of the defects of the prior art.

In order to achieve the above purpose, the present invention adopts the following technical solutions:

The channel estimation method in the FDD downlink multi-user massive MIMO system includes the steps:

S1. Design training sequences for multiple users and process received signals;

S2. Perform corresponding angle estimation according to the received signal;

S3. Construct a channel steering vector through the angle estimation value;

S4. Reconstruct downlink channel state information according to the constructed channel steering vector.

Further, in the step S1, multiple users design training sequences and process received signals, including K mutually independent single-antenna users, one base station, and M antennas at the base station.

Further, the M antennas equipped at the base station are arranged in a plane array, and M=M _h M _v , wherein each row has M _h antennas, and each column has M _v antennas, then the kth (k=1 , 2, 3, ..., K) users' channels from the base station to the receiver are specifically expressed as:

Where: P _k represents the number of propagation paths between the base station and the receiver of the kth user; gk _,l represents the channel complex gain of the lth path of the kth user; α(θ,φ) represents the channel steering vector; α(θ _k,l ,φk _,l ) represents the steering vector of the channel of the lth path of the kth user; θk _,l and φk _,l respectively represent the lth path of the kth user of elevation and azimuth.

Further, the steering vector α(θ,φ) of the channel is expressed as:

Among them: d represents the distance between the antenna elements, assuming d=λ/2, λ is the carrier wavelength; α _v (θ) contains the longitudinal angle information of the planar array, and α _h (θ, φ) contains the horizontal direction of the planar array. angle information.

Further, the training sequence designed in the step S1 is specifically:

It is assumed that the transmitted signals of the base station to all K users are S _N , which is an M×N-dimensional matrix, which is specifically expressed as:

Where: F _n represents an n-dimensional discrete Fourier DFT matrix; 0 _n is an n-dimensional zero matrix; 0 _(M-7n)×n is a (M-7n)×n-dimensional zero matrix, and n satisfies 4n=N , and N is the total number of samples.

Further, the processing of the received signal in the step S1 is specifically:

After mixing with noise, the received signal of the kth (k=1, 2, 3, .., K) user terminal is expressed as:

Among them: ρ is the signal-to-noise ratio, z _k is the noise mixed in the kth user, assuming that it is Gaussian noise that obeys zero mean and unit variance;

进行反DFT变换，则在第k个用户处的实际接收信号表示为：to send a signal

Perform inverse DFT transform, then the actual received signal at the kth user is expressed as:

where y _k represents the received signal.

Further, the step S2 specifically includes:

S21. Set the codebook, and regard the cosθsinφ in the horizontal vector α _h (θ, φ) in the channel steering vector α (θ, φ) and the sinθ in the longitudinal vector α _v (θ) as the interval in (-1, 1) two wholes; then create two m codebooks respectively in the (-1,1) interval, and m=M/2;

S22. The received signal of the _i (i=1, 2, .

和

表示每次遍历设定的角度码本的取值；

表示每次遍历设定的角度码本后的取值；in,

and

Indicates the value of the angle codebook set for each traversal;

Indicates the value after each traversal of the set angle codebook;

S23. In combination with the preset codebook, the codeword is set as follows:

遍历预先设定的码本，从y_k,r(i)中挑选出投影功率最大的码字，表示为：Combining set codewords

Traverse the preset codebook, and select the codeword with the largest projection power from y _k,r (i), which is expressed as:

和

为第k个用户的第i条路径中的信道导向矢量中的角度信息的估计值，利用得到的角度信息的估计值，重新弄构造信道的导向矢量

in the codeword

and

is the estimated value of the angle information in the channel steering vector in the ith path of the kth user, and reconstructs the channel steering vector using the obtained estimated value of the angle information

S24. Combine the preset codebook, traverse the angle codebook, select and obtain the estimated value of the angle information, so as to estimate the gain of the i-th path of the k-th user, which is expressed as:

Then the estimated value of the channel complex gain is:

Further, the step S3 is specifically:

通过信道导向矢量α(θ,φ)中的α_v(θ)和α_h(θ,φ)的形式构造出导向矩阵横向矢量

和纵向矢量

二者做张量积

构造出信道导向矢量

Combined with angle information estimates

The steering matrix transverse vector is constructed in the form of α _v (θ) and α _h (θ, φ) in the channel steering vector α (θ, φ).

and portrait vector

The two do tensor product

Construct the channel steering vector

以及信道复增益估计值

重构信道矩阵从而恢复下行信道信息。Further, the step S4 is specifically: according to the reconstructed channel steering vector

and the channel complex gain estimate

The channel matrix is reconstructed to restore downlink channel information.

Compared with the prior art, the present invention is based on the channel estimation method based on a small number of uniform training sequences, and multiple users can estimate the channel state information at the same time, effectively reducing pilot frequency overhead, and laying a foundation for realizing higher system throughput.

Description of drawings

1 is a flowchart of a channel estimation method in an FDD downlink multi-user massive MIMO system according to Embodiment 1;

FIG. 2 is a simulation diagram of a channel estimation method in an FDD downlink multi-user massive MIMO system according to the first embodiment.

Detailed ways

The embodiments of the present invention are described below through specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments may be combined with each other under the condition of no conflict.

Example 1

This embodiment provides a channel estimation method in an FDD downlink multi-user massive MIMO system, as shown in FIG. 1 , including:

S1. Design training sequences for multiple users and process received signals;

S2. Perform corresponding angle estimation according to the received signal;

S3. Construct a channel steering vector through the angle estimation value;

S4. Reconstruct downlink channel state information according to the constructed channel steering vector.

In the channel estimation method in the FDD downlink multi-user massive MIMO system in this embodiment, the channel estimation method based on a small number of uniform training sequences is used in the downlink multi-user massive MIMO system, aiming at the shortcomings of the existing channel estimation methods. Improvements have been made.

The specific application cases are as follows:

Assume that there is 1 cell and 1 base station in the system, the number of antennas of the base station is 128, the sampling times N is 32, n is 8, and m is 64. The number of paths of wireless signals from the base station to each user is P _k =6, and each user has one receiving antenna. The number of antennas in each row of the planar array antenna M _h = 16, the number of antennas in each column M _v = 8, the carrier wavelength λ = 3 × 10 ⁸ /1.5 × 10 ⁹ , the array element spacing d = λ/2, the channel matrix is guided The elevation angle θ and azimuth angle φ contained in the vector are uniformly distributed in the interval (-900,900].

In step S1, training sequences are designed for multiple users and received signals are processed.

The 128 antennas equipped at the base station are arranged in a planar array (UPA Uniform-Planar-Array), and M=16×8, in which there are 16 antennas in each row and 8 antennas in each column. The channel of the kth user from the base station to the receiver is specifically expressed as:

Among them: P _k is the number of propagation paths between the base station and the receiving end of the kth user, _gk,l represents the channel complex gain of the lth path of the kth user, assuming that it obeys the zero mean, the unit variance complex Gaussian distribution. α(θ _k,l , φ _k,l ) represents the steering vector of the channel of the lth path of the kth user, θk _,l and φk _,l respectively represent the elevation angle of the lth path of the kth user and azimuth.

The design training sequence is as follows:

For all K users, the basis for selecting the training sequence at the base station is: the sampling times of the horizontal angle information and the vertical angle information in the channel steering vector of the lth path of the kth user should satisfy the requirements of the planar array antenna. The ratio of the number of rows to the number of columns.

Assuming that the transmitted signals of the base station to all K users are S _N , which is a 128×32-dimensional matrix, which is specifically expressed as:

Where: F _n is an 8-dimensional discrete Fourier (DFT) matrix, 0 _n is an 8-dimensional zero matrix, 0 _(M-7n)×n is a 72×8-dimensional zero matrix, and n satisfies 4n=N, N is the total number of samples.

The processing of the received signal is as follows:

After mixing noise, the received signal of the kth user terminal is expressed as:

Among them: ρ is the signal-to-noise ratio, z _k is the noise mixed in at the kth user, assuming that it is Gaussian noise with zero mean and unit variance.

进行反DFT变换，则在第k个用户处的实际接收信号表示为：to send a signal

Perform inverse DFT transform, then the actual received signal at the kth user is expressed as:

where y _k represents the received signal.

In step S2, corresponding angle estimation is performed according to the received signal.

A codebook of angles is created in advance, and then all codebooks are traversed.

S21. Set the codebook, and regard the cosθsinφ in the horizontal vector α _h (θ, φ) in the channel steering vector α (θ, φ) and the sinθ in the longitudinal vector α _v (θ) as the interval in (-1, 1) of the two wholes. Then, two 64 codebooks are created in the (-1,1) interval.

S22. The received signal of the i (i=1, 2, 3, 4, 5, 6) path at the kth user is y _k (i), and the noise of the lth path of the kth user is deduced as follows :

和

表示每次遍历设定的角度码本的取值，

表示每次遍历设定的角度码本后的取值。in:

and

Indicates the value of the angle codebook set for each traversal,

Indicates the value after each traversal of the set angle codebook.

S23. In combination with the preset codebook, the codeword is set as follows:

遍历预先设定的码本，从y_k,r(i)中挑选出投影功率最大的码字，具体计算公式如下：Combining set codewords

Traverse the preset codebook, and select the codeword with the largest projection power from y _k,r (i). The specific calculation formula is as follows:

和

就是第k个用户的第i条路径中的信道导向矢量中的角度信息的估计值，利用得到的角度信息的估计值，就可以重新弄构造信道的导向矢量

in the codeword

and

is the estimated value of the angle information in the channel steering vector in the i-th path of the kth user. Using the obtained estimated value of the angle information, the channel steering vector can be reconstructed

S24. Combine the preset codebook, traverse the angle codebook, select and obtain the estimated value of the angle information, so as to estimate the gain of the i-th path of the k-th user, and the specific calculation formula is as follows:

Then the estimated value of the channel complex gain is:

In step S3, a channel steering vector is constructed from the angle estimates.

通过信道导向矢量α(θ,φ)中的α_v(θ)和α_h(θ,φ)的形式构造出导向矩阵横向矢量

和纵向矢量

二者做张量积

可构造出信道导向矢量

Combined with angle information estimates

The steering matrix transverse vector is constructed in the form of α _v (θ) and α _h (θ, φ) in the channel steering vector α (θ, φ).

and portrait vector

The two do tensor product

The channel steering vector can be constructed

In step S4, the downlink channel state information is reconstructed according to the constructed channel steering vector.

以及信道复增益估计值

重构信道矩阵从而恢复下行信道信息。According to the reconstructed channel steering vector

and the channel complex gain estimate

The channel matrix is reconstructed to restore downlink channel information.

In this embodiment, in order to analyze the performance of the proposed channel estimation method based on angle estimation in the FDD large-scale antenna system, the system throughput is defined as: C=log(1+ρ||h ^H f|| ² ) , f is precoding.

As shown in Figure 2, it is a simulation diagram of system throughput under the conditions of the above example, in which the "ideal transition" is the system throughput curve using the real channel for precoding, and the method proposed in this embodiment uses reconstruction System throughput curve for channel precoding. It can be seen from FIG. 2 that the precoding of the channel estimation value obtained by the method proposed in this embodiment is compared with the precoding of the real channel under the same conditions, and the difference in the throughput of the system is less than about 1 bit, and The signal-to-noise ratio has little influence on the method proposed by the present invention, and has good system performance.

Compared with the prior art, in this embodiment, in order to better save training overhead, after theoretical derivation, it can be deduced in combination with the PRONY equation. For a large-scale antenna system, for an array antenna, the channels are all in a form that satisfies the PRONY equation. Based on this, considering that in a multi-user scenario, different users can be trained by sending the same training sequence at the same time, and combined with the angle estimation method to obtain complete channel state information to reduce pilot overhead and time consumption. The channel estimation method based on a small number of uniform training sequences allows multiple users to estimate channel state information simultaneously, effectively reducing pilot overhead and laying the foundation for achieving higher system throughput.

The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which the present invention pertains can make various modifications or additions to the described specific embodiments or substitute in similar manners, but will not deviate from the spirit of the present invention or go beyond the definitions of the appended claims range.

Claims (9)

1. The channel estimation method in the FDD downlink multi-user massive MIMO system, is characterized in that, comprises the steps: S1. Design training sequences for multiple users and process received signals; S2. Perform corresponding angle estimation according to the received signal; S3. Construct a channel steering vector through the angle estimation value; S4. Reconstruct downlink channel state information according to the constructed channel steering vector. 2. The channel estimation method in the FDD downlink multi-user massive MIMO system according to claim 1, wherein, in the step S1, a plurality of users design training sequences and process the received signal, including K mutually independent single Antenna users, one base station, and M antennas are provided at the base station. 3. The channel estimation method in the FDD downlink multi-user massive MIMO system according to claim 2, wherein the M antennas equipped at the base station adopt a planar array arrangement, and M=M _h M _v , There are M _h antennas in each row and M _v antennas in each column, then the channel of the kth (k=1,2,3,...,K) user from the base station to the receiving end is specifically expressed as:

Where: P _k represents the number of propagation paths between the base station and the receiver of the kth user; gk _,l represents the channel complex gain of the lth path of the kth user; α(θ,φ) represents the channel steering vector; α(θ _k,l ,φ _k,l ) represents the steering vector of the channel of the lth path of the kth user; θk _,l and φk _,l respectively represent the lth path of the kth user of elevation and azimuth. 4. The channel estimation method in the FDD downlink multi-user massive MIMO system according to claim 3, wherein the steering vector α(θ, φ) of the channel is expressed as:

Among them: d represents the distance between the antenna elements, assuming d=λ/2, λ is the carrier wavelength; α _v (θ) contains the longitudinal angle information of the planar array, and α _h (θ, φ) contains the horizontal direction of the planar array. angle information. 5. The channel estimation method in the FDD downlink multi-user massive MIMO system according to claim 4, wherein the design training sequence in the step S1 is specifically: It is assumed that the transmitted signals of the base station to all K users are S _N , which is an M×N-dimensional matrix, which is specifically expressed as:

Where: F _n represents an n-dimensional discrete Fourier DFT matrix; 0 _n is an n-dimensional zero matrix; 0 _(M-7n)×n is a (M-7n)×n-dimensional zero matrix, and n satisfies 4n=N , and N is the total number of samples. 6. The channel estimation method in the FDD downlink multi-user massive MIMO system according to claim 5, wherein the processing of the received signal in the step S1 is specifically: After mixing with noise, the received signal of the kth (k=1, 2, 3, .., K) user terminal is expressed as:

Among them: ρ is the signal-to-noise ratio, z _k is the noise mixed in the kth user, assuming that it is Gaussian noise that obeys zero mean and unit variance; to send a signal

Perform inverse DFT transform, then the actual received signal at the kth user is expressed as:

where y _k represents the received signal. 7. The channel estimation method in the FDD downlink multi-user massive MIMO system according to claim 6, wherein the step S2 specifically comprises: S21. Set the codebook, and regard the cosθsinφ in the horizontal vector α _h (θ, φ) in the channel steering vector α (θ, φ) and the sinθ in the longitudinal vector α _v (θ) as the interval in (-1, 1) two wholes; then create two m codebooks respectively in the (-1,1) interval, and m=M/2;

S22. The received signal of the _i (i=1, 2, .

in,

and

Indicates the value of the angle codebook set for each traversal;

Indicates the value after each traversal of the set angle codebook; S23. In combination with the preset codebook, the codeword is set as follows:

Combining set codewords

Traverse the preset codebook, and select the codeword with the largest projection power from y _k,r (i), which is expressed as:

in the codeword

and

is the estimated value of the angle information in the channel steering vector in the ith path of the kth user, and reconstructs the channel steering vector using the obtained estimated value of the angle information

S24. Combine the preset codebook, traverse the angle codebook, select and obtain the estimated value of the angle information, so as to estimate the gain of the i-th path of the k-th user, which is expressed as:

Then the estimated value of the channel complex gain is:

8. The channel estimation method in the FDD downlink multi-user massive MIMO system according to claim 7, wherein the step S3 is specifically: Combined with angle information estimates

The steering matrix transverse vector is constructed in the form of α _v (θ) and α _h (θ, φ) in the channel steering vector α (θ, φ).

and portrait vector

The two do tensor product

Construct the channel steering vector

9. The channel estimation method in the FDD downlink multi-user massive MIMO system according to claim 8, wherein the step S4 is specifically: according to the reconstructed channel steering vector

and the channel complex gain estimate

The channel matrix is reconstructed to restore downlink channel information.

Images