阅读说明：本技术 一种基于训练序列的宽带混合波束赋形方法 (Broadband mixed beam forming method based on training sequence ) 是由 王�华 孙艺玮 何东轩 袁明浩 朱田野 于 2021-07-26 设计创作，主要内容包括：本发明针对多用户大规模MIMO-OFDM时分双工系统,提出一种基于训练序列的宽带混合波束赋形方法。首先基于最小均方误差准则,利用接收信号的自相关以及发送和接收信号的互相关,通过上行和下行的迭代过程,分别设计下行和上行用户端全数字波束赋形器；其次,利用主成分分析法提取频率平坦的射频波束赋形器；然后在所述射频波束赋形器基础上对其上行等效基带信道进行信道估计；最后根据所述信道估计结果采用块对角化的方法对基带波束进行赋形。(The invention provides a broadband mixed beam forming method based on a training sequence aiming at a multi-user large-scale MIMO-OFDM time division duplex system. Firstly, based on the minimum mean square error criterion, respectively designing downlink and uplink user end all-digital beam formers by utilizing autocorrelation of received signals and cross-correlation of transmitted and received signals and through uplink and downlink iterative processes; secondly, extracting a radio frequency beam former with flat frequency by using a principal component analysis method; then, channel estimation is carried out on the uplink equivalent baseband channel of the radio frequency beam former on the basis of the radio frequency beam former; and finally, forming a baseband beam by adopting a block diagonalization method according to the channel estimation result.)

1. A broadband mixed beam forming method based on training sequence is characterized in that firstly, based on the minimum mean square error criterion, the self-correlation of received signals and the cross-correlation of transmitted and received signals are utilized, and through the uplink and downlink iterative processes, the full digital beam forming devices of downlink and uplink user terminals are respectively designed; secondly, extracting a radio frequency beam former with flat frequency by using a principal component analysis method; then, channel estimation is carried out on the uplink equivalent baseband channel of the radio frequency beam former on the basis of the radio frequency beam former; and finally, forming a baseband beam by adopting a block diagonalization method according to the channel estimation result.

2. The wideband hybrid beamforming method based on training sequence according to claim 1, wherein the all-digital beamformer of the downlink and uplink user terminals is specifically designed in the following manner:

step 1.1, initialization

Setting the base station full-digital beam former asThe maximum iteration number is I, and an iteration number counter I is 0;

step 1.2, designing all-digital beam shaping device of downlink user terminal

According to the minimum mean square error criterion, the all-digital beam former of the downlink user terminal is designed as follows:

in which the base station transmits a training signal s^DL[k]，The received signal at the u-th user antenna end isIn which additive white Gaussian noiseSigma is the noise standard deviation),is an autocorrelation matrix of the received signal,a cross-correlation matrix for the received signal and the transmitted signal;

step 1.3, designing the all-digital beam former of the uplink base station end

According to the minimum mean square error criterion, designing an all-digital beam former at an uplink base station end as follows:

F^MMSE[k]＝E[r^UL[k]r^ULH[k]]^-1E[r^UL[k]s^ULH[k]]；

wherein the u-th user transmits a training signal For two different users i and j there isThe received signal at the antenna end of the base station is

Step 1.4, judging iteration termination conditions

if the I is greater than or equal to I +1, judging whether the I is greater than or equal to I, if so, carrying out the next step, otherwise, returning to the step 1.2;

and step 1.5, extracting the radio frequency beam former with flat frequency by using a principal component analysis method.

3. The method as claimed in claim 2, wherein the method comprises: the method for extracting the radio frequency beam shaper with flat frequency by using the principal component analysis method specifically comprises the following steps:

at the base station side, a data set matrix F ═ F is defined^MMSE[1]…F^MMSE[K]]In which F is^MMSE[k]For all digital beam-formers under each subcarrier obtained in step 1.3, singular value decomposition is performed on the data set matrixThe beam shaper at the base station end is

Wherein L is_BSIs the number of radio frequency chains at the base station end, N_BSIs the number of base station antennas, [ A ]]_:,1:BRepresenting to take the first B column of the matrix A, taking the argument of each element of the complex matrix, taking j as an imaginary unit, and exp (DEG) representing an exponential function with a natural constant e as a base;

at the u-th user terminal, defining a data set matrix Performing singular value decomposition on the full digital beam former under each subcarrier obtained in the step 1.2The base station beamformer isL_MSNumber of radio frequency chains for user terminal, N_MSThe number of user antennas.

4. A method as claimed in claim 1, 2 or 3, wherein the method comprises: performing channel estimation on an uplink equivalent baseband channel of the radio frequency beam former on the basis of the radio frequency beam former, specifically:

let the transmitted training signal of the u-th user be x_u[k]Then the received signal of the base station isWhereinAccording to the least square principle, the estimation result of the baseband equivalent channel is

5. A method as claimed in claim 1, 2 or 3, wherein the method comprises: designing baseband beamforming, which comprises the following steps:

step 3.1, calculating an inter-user interference suppression matrix;

record the aggregate matrix of the u user interference channel as

Its singular value decomposition intoLet L_MSThe matrix formed by the right singular vectors corresponding to the minimum singular value isMatrix arrayNamely an inter-user interference suppression matrix;

3.2, eliminating the interference among users of the equivalent channel and carrying out singular value decomposition:

eliminating the u-th user interferenceTo pairPerforming singular value decomposition

Step 3.3, designing a baseband beam former;

the base band beam shaper of the U-th user served by the base station end isThe u-th user end base band beam shaper is

Technical Field

The invention relates to the field of beam forming in wireless communication, and mainly considers that in a multi-user time-division duplex (TDD) millimeter wave large-scale MIMO orthogonal frequency division multiple access (OFDM) communication system, a low-complexity channel estimation technology is combined, and a base station and a user respectively eliminate interference among users and interference among different data streams of the same user so as to improve communication capacity.

Background

With the rapid development of technologies such as the internet of things and the mobile internet, the requirements of social life on the speed and reliability of wireless data transmission are greatly improved, and the millimeter wave large-scale MIMO technology is one of key technologies for realizing high-speed data transmission. Meanwhile, the wavelength of the millimeter wave frequency band is extremely short, and the integration and miniaturization of the large-scale antenna array are easy to realize. Therefore, the millimeter wave large-scale MIMO technology is also one of the key technologies for meeting the communication requirements of aircrafts such as unmanned planes, satellites and the like. In a millimeter wave large-scale MIMO communication system, the mixed beam forming can obviously improve the spectrum efficiency on the basis of reducing the energy consumption. An efficient hybrid beamforming scheme in a multi-user environment is highly dependent on accurate channel state information. However, for multi-user massive MIMO with tens or hundreds of antenna elements, a high-precision low-complexity channel estimation method is challenging. Therefore, a hybrid beamforming scheme operating under imperfect channel state information or combined with channel estimation is a considerable problem.

Multi-user hybrid beamforming schemes utilizing imperfect channel state information and multi-user hybrid beamforming schemes combined with channel estimation have been widely studied. For example: jos ep p.gonz-lez-Coma proposes a hybrid beamforming design for multiuser systems with incomplete channel information in frequency selective channels. The scheme is based on the autocorrelation of a received signal and the cross-correlation estimation of the received signal and a transmitted signal, and adopts an iterative Minimum Mean Square Error (MMSE) algorithm to design a hybrid beamformer.

Aiming at different requirements and flexibility of a radio frequency and baseband beamforming device on channel state information on a time scale, Yinglei Teng provides a mixed time scale mixed beamforming scheme under a self-adaptive partial connection radio frequency beamforming structure. To reduce feedback overhead, Chunmei Xu designs a hybrid precoding scheme using a given analog codebook and corresponding effective channel state information. Jos better P, Gonz & lez-Coma combines with a synchronous weighted orthogonal matching pursuit (SW-OMP) channel estimation method, and proposes a projection gradient-based hybrid beamforming scheme. The Ahmed Alkhateeb introduces the most advanced deep learning technology and develops a beam forming solution supporting high-mobility millimeter wave application. In addition, An-An Lu studies the design of robust linear beamformers for massive MIMO downlink in case the channel state information is incomplete.

However, despite the many hybrid beamforming schemes proposed, most hybrid beamforming designs have limitations in their research. Although some of the documents design the hybrid beamformer with imperfect channel state information, they do not propose detailed methods of acquiring statistical or effective channel state information. On the other hand, some documents combine channel estimation with hybrid beamforming design, but they typically use very sparse multipath models, which may not have satisfactory performance in rich multipath situations. Therefore, new schemes need to be explored to reduce the channel state information requirement of hybrid beamforming and provide a low-complexity estimation method of incomplete channel state information.

Disclosure of Invention

In view of this, the present invention provides a wideband hybrid beamforming method based on a training sequence for a multi-user large-scale MIMO-OFDM time division duplex system.

The invention is realized by the following technical scheme.

A broadband mixed beam forming method based on training sequence, based on the minimum mean square error criterion, utilize autocorrelation of the received signal and cross-correlation of the transmitted and received signal, through the up run and downstream iterative process, design the all-digital beam forming device of the downstream and upstream user end separately; secondly, extracting a radio frequency beam former with flat frequency by using a principal component analysis method; then, channel estimation is carried out on the uplink equivalent baseband channel of the radio frequency beam former on the basis of the radio frequency beam former; and finally, forming a baseband beam by adopting a block diagonalization method according to the channel estimation result.

The invention has the beneficial effects that:

(1) the invention adopts the autocorrelation of the received signal and the cross correlation of the received signal and the transmitted signal, and resists the frequency domain selectivity of the broadband channel by the uplink and downlink iterative design and the principal component analysis method, so that the radio frequency beam shaping device can be designed under the condition of unknown channel state information, and the radio frequency beam shaping of the base station and the user can be completed locally, thereby saving the feedback link of the radio frequency beam shaping and the feedback expense.

(2) The invention adopts digital domain equivalent channel estimation, and the channel estimation algorithm can be used under more physical channel models, thereby reducing the complexity of channel estimation.

Drawings

Fig. 1 is a flowchart of a wideband hybrid beamforming method based on a training sequence according to the present invention;

fig. 2 is a schematic diagram of modeling a channel communication scenario system in the embodiment of the present invention.

Detailed Description

The invention will be further described with reference to the accompanying drawings.

As shown in FIG. 2, the embodiment is channel communication scenario system modeling in a multi-user massive MIMO-OFDM system based on TDD, wherein a base station is equipped with N_BSRoot antenna and L_BSAnd the radio frequency chains meet the condition that the radio frequency quantity is far less than the quantity of the antennas. In addition, the base station can serve a maximum of U users simultaneously, with each user being equipped with N_MSRoot antenna and L_MSThe number of the radio frequency chains is far less than that of the antennas. Each user can support N_sOne data stream is transmitted, then the base station can support UN_sTransmitting data stream, and using each radio frequency chain to maximum extent, wherein the number of data stream and radio frequency chain is assumed to satisfy N_s＝L_MSAnd UN_s＝L_BS. The number of subcarriers of the OFDM system is K.

The millimeter wave channel is assumed to be composed of N_clSum of contributions of scattering clusters, wherein each cluster contains N_rayWhen there is a path and each path has a corresponding timeDelay tau_u,l,rThus, the channel expression from the base station to the u user at the d-th time delay can be written as follows

Wherein alpha is_u,l,rDenotes the complex gain of the r-th path in the l-th cluster, and a_u,l,rObeying a complex Gaussian distribution CN (0, 1); p is a radical of_rc,u(τ) is a shaping filter, the signal period is T_s；a_BSAnd a_MS,uRespectively representing array response vectors of a base station antenna and a user antenna, considering that the antennas are uniform linear arrays with the antenna spacing of carrier intermediate frequency half-wavelength, and the array correspondence can be represented as

Wherein, theta_BSAnd theta_MS,uRespectively, angle of departure and angle of arrival. The frequency domain channel matrix may be further written as

Where D is the length of the cyclic prefix and the diagonal matrix Δ_u[k]Comprising a complex gain per path, A_BSAnd A_MS,uIs a matrix composed of the corresponding vectors of the array of all the paths.

By utilizing the channel reciprocity of the TDD system, the system is assumed to share the same beam forming scheme in the uplink and downlink processes.

Based on the modeling analysis, a method for forming a wideband hybrid beam based on a training sequence according to this embodiment is further described with reference to fig. 1, which specifically includes the following steps:

step one, radio frequency beam forming design

In the embodiment, the design of the radio frequency beam forming is based on that under the condition that channel state information is unknown, firstly, based on the minimum mean square error criterion, a downlink user end and an uplink user end all-digital beam forming device are respectively designed by utilizing autocorrelation of received signals and cross-correlation of transmitted and received signals through an uplink iterative process and a downlink iterative process; then, a frequency-flat RF beamformer F is extracted by a principal component analysis method_RFAnd W_RF,u. The method specifically comprises the following steps:

step 1.1, initialization

Setting the base station full-digital beam former asThe maximum iteration number is I, and an iteration number counter I is 0;

step 1.2, designing all-digital beam shaping device of downlink user terminal

According to the minimum mean square error criterion, the all-digital beam former of the downlink user terminal is designed as follows:

in which the base station transmits a training signal s^DL[k]，The received signal at the u-th user antenna end is(additive white Gaussian noise)Sigma is the noise standard deviation),is an autocorrelation matrix of the received signal,a cross-correlation matrix for the received signal and the transmitted signal;

step 1.3, designing the all-digital beam former of the uplink base station end

According to the minimum mean square error criterion, designing an all-digital beam former at an uplink base station end as follows:

F^MMSE[k]＝E[r^UL[k]r^ULH[k]]^-1E[r^UL[k]s^ULH[k]]，

wherein the u-th user transmits a training signalFor two different users i and j there isThe received signal at the antenna end of the base station is(additive white Gaussian noise)Sigma is the standard deviation of noise), E [ r^UL[k]r^ULH[k]]For the autocorrelation matrix of the received signal, er^UL[k]s^ULH[k]]A cross-correlation matrix for the received signal and the transmitted signal;

step 1.4, judging iteration termination conditions

if the I is greater than or equal to I +1, judging whether the I is greater than or equal to I, if so, carrying out the next step, otherwise, returning to the step 1.2;

step 1.5, extracting the flat-frequency radio frequency beam former by using a principal component analysis method

In the present embodiment, the following method is adopted to extract the RF beamformer F with flat frequency_RFAnd W_RF,u：

At the base station side, a data set matrix F ═ F is defined^MMSE[1]…F^MMSE[K]](F^MMSE[k]All-digital beamformer under each subcarrier obtained in step 1.3), odd for the dataset matrixSingular value decompositionThe beam shaper at the base station end is(L_BSIs the number of radio frequency chains at the base station end, N_BSIs the number of base station antennas, [ A ]]_:,1:BRepresenting to take the first B column of the matrix A, taking the argument of each element of the complex matrix, taking j as an imaginary unit, and exp (a) representing an exponential function with a natural constant e as a base); at the u-th user terminal, defining a data set matrix(All-digital beam-former under each subcarrier obtained in step 1.2), singular value decomposition is carried out on the full-digital beam-formerThe base station beamformer is(L_MSNumber of radio frequency chains for user terminal, N_MSNumber of user antennas).

Step two, channel estimation is carried out on the uplink equivalent baseband channel

In this embodiment, channel estimation is performed on the uplink equivalent baseband channel of the rf beamformer based on the rf beamformer, that is, F is designed in step one_RFAnd W_RF,uOn the basis of, toAnd performing channel estimation to obtain a channel estimation result.

The method specifically comprises the following steps: let the transmitted training signal of the u-th user be x_u[k]Then the received signal of the base station isWhereinAccording to the least square principle, the estimation result of the equivalent channel of the baseband is obtained

Third step, base band beam forming design

According to the base band equivalent channel estimation result H_eq,u[k]The method of block diagonalization is adopted to carry out beamforming on the basebandAndand (5) designing. The method comprises the following specific steps:

step 3.1, calculating an inter-user interference suppression matrix

Record the aggregate matrix of the u user interference channel as

Its singular value decomposition intoLet L_MSThe matrix formed by the right singular vectors corresponding to the minimum singular value isMatrix arrayI.e. the inter-user interference suppression matrix.

3.2, eliminating the interference among users of the equivalent channel and carrying out singular value decomposition:

eliminating the u-th user interferenceTo pairPerforming singular value decomposition

Step 3.3, designing a baseband beam former:

the base band beam shaper of the U-th user served by the base station end isThe u-th user end base band beam shaper is

In specific implementation, for a time-varying channel, the present embodiment adopts a mode of rf beamforming lasting for a longer channel coherence time, and baseband beamforming lasting for a shorter channel coherence time. For the limited feedback system, the rf beamforming design of this embodiment can be completed locally at the base station and the user end, respectively, and thus no feedback is required. In order to reduce the feedback overhead, a feedback scheme is adopted in which baseband beamforming is designed by a base station and the design result of the baseband beamforming of a user is fed back to the user by the base station.

In summary, the above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.