CN110752866B

CN110752866B - Large-scale MIMO precoding transmission method and device

Info

Publication number: CN110752866B
Application number: CN201810816736.4A
Authority: CN
Inventors: 高西奇; 郭伟然; 卢安安; 孟鑫; 马霓
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2018-07-23
Filing date: 2018-07-23
Publication date: 2021-08-20
Anticipated expiration: 2038-07-23
Also published as: CN110752866A

Abstract

The embodiment of the application discloses a large-scale multiple-input multiple-output (MIMO) precoding transmission method, which is used for realizing wide-angle coverage of transmitting power of transmitting signals in a large-scale MIMO system and improving the utilization efficiency of the power. The method in the embodiment of the application comprises the following steps: the access network equipment determines a target function according to a preset target wide coverage power pattern; the access network equipment processes the precoding matrix corresponding to the objective function by using a steepest descent method to obtain a wide coverage precoding matrix, a power pattern generated by the wide coverage precoding matrix approaches to a target wide coverage power pattern, the wide coverage precoding matrix is an MxN dimensional precoding matrix, and M and N are integers larger than 0; the access network equipment multiplies the N-dimensional signal by a wide coverage pre-coding matrix to obtain an M-dimensional transmitting signal, and the N-dimensional signal is generated by the access network equipment; the access network equipment sends an M-dimensional transmitting signal, and the wide-coverage precoding matrix is a precoding matrix of transmitting power wide-angle coverage of the transmitting signal.

Description

Large-scale MIMO precoding transmission method and device

Technical Field

The present application relates to the field of wireless communication technologies, and in particular, to a large-scale MIMO wide-coverage precoding transmission method and apparatus.

Background

A large-scale multiple-input multiple-output (multiple-output) technology that a large-scale antenna array is equipped at a Base Station (BS) to greatly improve system capacity is one of key technologies of a new generation of wireless networks in the future, and is also a research hotspot in recent years. One of the challenging technical difficulties of large-scale multiple-input multiple-output (multiple-input multiple-output) technology is to design the signaling process for the common channel. Currently, there is relatively little research on common channel transmission techniques. Common channels play an important role in cellular systems, and much of the necessary common information and services are delivered to users over common channels. Since the common channel serves all users in the cell and not only certain active users, it is desirable that the radiation power of the signals transmitted by the base station hardly fluctuates in different spatial directions in the cell so that all users in the cell can reliably receive the common signal.

At present, there is a 180-degree omni-directional coverage scheme for transmitting signals, where the radiation power of signals transmitted in the 180-degree omni-directional coverage scheme can meet the requirement of almost no fluctuation in different spatial directions in a cell, and the technical scheme uses a Zadoff-Chu (ZC for short) sequence, which is a sequence sent by a communication signal, and is widely used in a Long Term Evolution (LTE) system due to its good auto-correlation and cross-correlation. The steps of the technical scheme are as follows: the access network equipment directly sets the power of certain angles to 1 (namely 0dB) through the omnidirectional antenna; then the access network equipment makes use of a ZC sequence to construct an M multiplied by N dimension precoding matrix, makes the power pattern of the precoding matrix keep constant in certain angles, the value is 0dB, makes each antenna unit in the omnidirectional antenna array realize equal power, and/or makes the precoding matrix containing the ZC sequence changed into a semi-unitary matrix, thereby maximizing the reachable traversal rate under an independent same distribution channel, wherein M and N are integers larger than 0, M is the number of antennas, and N is a signal generated by the access network equipment; then the access network equipment multiplies the N-dimensional signal by the precoding matrix to obtain an M-dimensional transmitting signal; and the access network equipment sends the M-dimensional transmitting signal.

However, the application of this scheme to a massive MIMO system will result in inefficient power amplifier utilization, because the 180-degree omni-directional coverage scheme can be implemented by single antenna transmission signals, and the total transmission power of the base station is often limited by the single power capacity of each antenna, so the power utilization efficiency needs to be further improved.

Disclosure of Invention

The embodiment of the application provides a large-scale MIMO precoding transmission method and access network equipment, which can realize wide-angle coverage of transmitting power of transmitting signals in a large-scale MIMO system and improve the utilization efficiency of the power.

A first aspect of the embodiments of the present application provides a large-scale MIMO precoding transmission method, including:

in order to realize the wide-angle coverage of power for transmitting signals by the access network equipment, the access network equipment may preset a target wide coverage power pattern, and then determine an objective function according to the target wide coverage power pattern, wherein the independent variable of the objective function is a M × N-dimensional precoding matrix, and therefore when the M × N-dimensional precoding matrix changes, the objective function also changes; then the access network equipment processes the precoding matrix corresponding to the objective function by using a steepest descent method to obtain a wide-coverage precoding matrix, so that the element values in the wide-coverage precoding matrix are continuously changed, so that the power pattern generated by the wide coverage precoding matrix is constantly close to the target wide coverage power pattern, i.e. the target function is constantly approaching 0, the wide coverage pre-coding matrix is a pre-coding matrix with dimension of M multiplied by N, the access network equipment pre-multiplies the pre-coding matrix with dimension of M multiplied by N to obtain a transmitting signal with dimension of M, where M is the number of antennas of the access network device, the N-dimensional signal is generated by the access network device, that is, the number of columns of the precoding matrix is not limited to 1, the access network device may have multiple data streams, and finally, and the access network equipment sends the M-dimensional transmitting signal, and the wide-coverage precoding matrix is a precoding matrix covered by the transmitting signal in a wide angle.

According to the technical scheme, the embodiment of the application has the following advantages: the access network equipment can determine the target function according to a preset target wide coverage power pattern, and then process the precoding matrix corresponding to the target function by using the steepest descent method to obtain the wide coverage precoding matrix, so that the power pattern generated by the wide coverage precoding matrix approaches to the target wide coverage power pattern, wide-angle coverage of the transmitting power when the large-scale MIMO system transmits the transmitting signals on each transmitting antenna can be realized, and the utilization efficiency of the transmitting power of the transmitting signals is improved.

Based on the first aspect of the present application, in a first implementation manner of the first aspect of the present application, the sending, by the access network device, the M-dimensional transmission signal includes:

the access network device sends the M-dimensional transmitting signals on each transmitting antenna, wherein the power of the M-dimensional transmitting signals sent on each transmitting antenna is the same, so that the power efficiency of each radio frequency channel of the M-dimensional transmitting signals and the power efficiency of the antenna array can reach the maximum value.

In the embodiment of the application, the access network device can enable the precoding matrix to meet the constraint condition of equal power through the steepest descent method while realizing wide-angle coverage of the power pattern, which means that the efficiency of the transmission power can be further greatly improved.

Based on the first implementation manner of the first aspect of the embodiment of the present application, the access network device may use a first formula as a constraint condition to make the M-dimensional transmission signals identical:

the first formula is as follows:

wherein W is the wide coverage precoding matrix, W^HFor the conjugate transpose of the wide-coverage precoding matrix, I_MIs an identity matrix with dimension of M rows and M columns "

"is the sign of the hadamard product. Through the formula, the wide-coverage precoding matrix is multiplied by the conjugate transpose of the wide-coverage precoding matrix, and the obtained matrix values on the diagonal line are equal, so that the access network equipment can enable the power of the M-dimensional transmitting signals to be the same.

In the embodiment of the application, the access network equipment can enable the power of the M-dimensional transmission signals to be the same through a specific constraint condition formula, so that the realizability of the scheme is improved in practical application.

Based on the second implementation manner of the first aspect of the embodiment of the present application, in the third implementation manner of the first aspect of the embodiment of the present application, the wide coverage precoding matrix belongs to a diagonal manifold, that is, an oblique manifold, and a characteristic of the diagonal manifold is that powers of transmission signals (that is, M-dimensional transmission signals) on M antennas can be made the same, so that power efficiencies of respective radio frequency channels of the M-dimensional transmission signals and the antenna array reach a maximum value, that is, the access network device sets the M-dimensional transmission signals to be the diagonal manifold.

In the embodiment of the application, the specific manifold is adopted to optimize the wide coverage precoding matrix, so that the M-dimensional transmitting signals have equal power, the scheme is further optimized, and the practicability of the scheme is improved.

Based on the first aspect of the embodiment of the present application or any one implementation manner of the first aspect to the third implementation manner of the first aspect of the embodiment of the present application, in a fourth implementation manner of the first aspect of the embodiment of the present application, the access network device may enable the wide-coverage precoding matrix to satisfy the following constraint condition:

the wide-coverage precoding matrix is changed into a semi-unitary matrix, so that the reachable traversal rate under independent same-distribution channels can be maximized, the reachable traversal rate is related to parameters such as the number of transmitting antennas, the number of receiving antennas, the transmitting power and the like, and the reachable traversal rate can be changed along with the value change of the parameters. When the access network device satisfies formula two, the wide coverage precoding matrix may be changed into a semi-unitary matrix:

W^HW＝I_N，

wherein, I_NIs an identity matrix with dimension of N rows and N columns.

In the embodiment of the application, the wide-coverage precoding matrix can be changed into the semi-unitary matrix according to a specific formula, so that the realizability of the scheme is improved.

Based on the first aspect of the present application or any one of the first implementation manner to the fifth implementation manner of the first aspect of the present application, in a sixth implementation manner of the first aspect of the present application, the objective function may be expressed as a function of a cosine distance between a power pattern generated by the wide coverage precoding matrix and the target wide coverage power pattern, and a formula of the objective function is as the following formula three:

the third formula is:

wherein the content of the first and second substances,

covering the power pattern for the target with a power pattern,

a power pattern generated for the wide coverage precoding matrix W,

is composed of

The transposing of (1).

In the embodiment of the application, the specific formula is provided for calculating the objective function, so that the realizability of the scheme is improved.

Based on the first aspect of the present application or any one of the first to fifth implementation manners of the first aspect of the present application, in a seventh implementation manner of the first aspect of the present application, the objective function may be expressed as a function of a cosine distance between an opening of a power pattern generated by the precoding matrix and an opening of the target wide coverage power pattern, and the formula of the objective function is as follows:

the fourth formula is:

wherein the content of the first and second substances,

is composed of

The square of the transpose of (a),

is prepared by

The method comprises the following steps of (1) making a square,

is composed of

The development of (1).

In the embodiment of the application, another specific formula is provided for calculating the objective function, so that the realizability and the flexibility of the scheme are further improved.

Based on the sixth implementation manner or the seventh implementation manner of the first aspect of the embodiment of the present application, in the eighth implementation manner of the first aspect of the present application, the target wide-coverage power pattern may be a KM-dimensional vector that has preset values of elements in the wide-coverage precoding matrix, where K is an oversampling coefficient, the power pattern generated by the wide-coverage precoding matrix is a KM-dimensional vector, and the power pattern generated by the wide-coverage precoding matrix may be a function of the precoding matrix and a transformation matrix.

In the embodiment of the present application, since the scheme is a specific scheme for presetting a target wide coverage power pattern and describes a power pattern generated by a specific wide coverage precoding matrix, the realizability of the scheme is improved.

Based on the eighth implementation manner of the first aspect of the embodiment of the present application, in a ninth implementation manner of the first aspect of the present application, the generating a power pattern by the wide coverage precoding matrix as a KM-dimensional vector includes:

the access network device may generate the KM-dimensional vector according to formula five, that is, the access network device may cause the wide-coverage precoding matrix to generate a power pattern of the KM-dimensional vector according to formula five:

the fifth formula is:

wherein the content of the first and second substances,

is a tangent space of the wide-coverage precoding matrix with dimension KM rows and M columns,

the dimension of KM rows and M columns is 2 × 8 rows and 8 columns, assuming that the oversampling factor is 2 and M is 8, which is the conjugate transpose of the tangent space of the wide coverage precoding matrix whose dimension is KM rows and M columns.

In the embodiment of the application, the power pattern generated by the wide coverage precoding matrix can be calculated through a specific formula, so that the practicability of the scheme can be further improved.

Based on the eighth implementation manner of the first aspect of the present application, in a tenth implementation manner of the first aspect of the present application, the transformation matrix may be a function of a steering vector and a directional pattern of a transmitting antenna of the access network device, and a formula of the transformation matrix may be the following formula six:

the sixth formula is:

wherein the content of the first and second substances,

represents the steering vector, | e (θ) | represents the directional pattern of the transmitting antenna.

In the embodiment of the application, the conversion matrix can be calculated through a specific formula, so that the practicability of the scheme is further improved.

Based on the sixth implementation manner or the seventh implementation manner of the first aspect of the example of the present application, in the eleventh implementation manner of the first aspect of the example of the present application, the steepest descent method may include the following steps:

the access network device may set a search direction only according to a gradient on the diagonal manifold, where the search direction is an inverse number of the gradient on the diagonal manifold; the access network device may also set the search direction only according to the gradient on the stefel manifold, in which case the search direction is set by the gradient on the stefel manifold; the access network device may further set a search direction according to a gradient at an intersection of the oblique manifold and the stefel manifold, where the search direction is an opposite number of the gradient at the intersection of the oblique manifold and the stefel manifold, and the search direction is a direction in which the search target function approaches zero;

the access network device sets the step length, for example, the access network device may set the step length according to an armijo linear search;

the access network equipment builds an updating equation by using the searching direction and the step length, wherein the updating equation is used for updating the precoding matrix into an updated precoding matrix;

the access network equipment judges whether a matrix difference value between the updated precoding matrix and the precoding matrix is smaller than a matrix preset threshold value or not;

if the matrix difference between the updated precoding matrix and the precoding matrix is smaller than a preset matrix threshold, the access network device may determine that the updated precoding matrix converges to the target precoding matrix.

In the embodiment of the present application, since the search direction may be set by one or two of the oblique manifold and the stefel manifold by using the above specific steepest descent method, and the updated precoding matrix is converged to the target precoding matrix according to the search direction and the set step length, so that the power pattern generated by the wide coverage precoding matrix is close to or even coincides with the power pattern generated by the target precoding matrix, thereby achieving wide coverage of the transmission power while achieving the equal power and the maximum value on each antenna, and improving the practicability of improving the power efficiency.

Based on the eleventh implementation manner of the first aspect of the embodiment of the present application, in the twelfth implementation manner of the first aspect of the embodiment of the present application, after the access network device determines whether a matrix difference between the updated precoding matrix and the precoding matrix is smaller than a preset matrix threshold, before the access network device sets a search direction according to a gradient on the oblique manifold or the shitifer manifold, or sets a search direction according to a gradient on an intersection of the oblique manifold and the shitifer manifold, the method further includes:

if the matrix difference between the updated precoding matrix and the precoding matrix is greater than or equal to a preset matrix threshold, the access network device may determine that the updated precoding matrix is not converged, that is, if the updated precoding matrix is not converged, the access network device may return to the step of performing the setting of the search direction again.

In the embodiment of the application, the scheme is provided when the updated precoding matrix is not converged, namely, the scheme is executed from the step of setting the search direction again, so that the realizability of the scheme is improved.

Based on the eleventh implementation manner of the first aspect of the present application, in a thirteenth implementation manner of the first aspect of the present application, the setting, by the access network device, a search direction according to a gradient on the diagonal manifold or the stifel manifold, or setting a search direction according to a gradient on an intersection of the diagonal manifold and the stifel manifold may include:

the access network device calculating a euclidean gradient of the objective function;

the access network equipment alternately projects the Euclidean gradient of the target function to the tangent space of the oblique manifold and the tangent space of the Stefel manifold, or projects the Euclidean gradient to the tangent space of the oblique manifold, or projects the Euclidean gradient to the tangent space of the Stefel manifold;

the access network equipment judges whether a first difference value between a current first projection result and a previous first projection result is smaller than a first preset threshold value or not;

if the access network device determines that the first difference is smaller than a first preset threshold, the access network device may determine that the sequence of the first projection converges to a gradient at an intersection of an oblique manifold and a stifel manifold, or a gradient of an oblique manifold, or a gradient of a stifel manifold, it should be noted that, if an euclidean gradient of the objective function is projected alternately to a tangent space of an oblique manifold and a tangent space of a stifel manifold, correspondingly, when the first difference is smaller than the first preset threshold, the sequence of the first projection may be determined to converge to a gradient at an intersection of an oblique manifold and a stifel manifold; if only the diagonal manifold is used, it may correspondingly be determined that the sequence of first projections converges to the diagonal manifold when the first difference is smaller than a first preset threshold; if only the stille manifold is used, it may correspondingly be determined that the sequence of first projections converges to the stille manifold when the first difference is less than a first preset threshold.

In the embodiment of the application, a specific scheme for setting the search direction by using the steepest descent method is provided, so that the equal power of the M-dimensional antenna can be specifically realized and the maximum value is reached, and the realizability of the scheme is improved.

Based on the thirteenth implementation manner of the first aspect of the embodiment of the present application, in a fourteenth implementation manner of the first aspect of the present application, after the access network device determines whether a first difference between a result of the current first projection and a result of the previous first projection is smaller than a first preset threshold, before the access network device calculates a euclidean gradient of the objective function, the method further includes:

if the access network device determines that the first difference is greater than or equal to a first preset threshold, the access network device may determine that the sequence of the first projection is not converged, and the access network device may return to the step of calculating the euclidean gradient of the objective function.

In the embodiment of the present application, since a scheme is provided when the sequence of the first projection is not converged, that is, the scheme is executed from the step of calculating the euclidean gradient of the objective function again, the realizability of the scheme is improved.

Based on the eleventh implementation manner of the first aspect of the embodiment of the present application, in a fifteenth implementation manner of the embodiment of the present application, the constructing, by the access network device, an update equation by using the search direction and the step size includes:

the access network equipment sums the product of the precoding matrix, the search direction and the step length to obtain a first numerical value;

the access network equipment alternately projects the first numerical value to the oblique manifold and the Stefel manifold, or projects the first numerical value to the oblique manifold, or projects the first numerical value to the Stefel manifold;

the access network equipment judges whether a second difference value between a current second projection result and a previous second projection result is smaller than a second preset threshold value or not;

if the access network device determines that the second difference is smaller than a second preset threshold, the access network device determines that the sequence of the second projection converges to the intersection of the diagonal manifold and the stefel manifold, or the diagonal manifold, or the stefel manifold, it should be noted that if the first value is projected alternately to the diagonal manifold and the stefel manifold, when the second difference is smaller than the second preset threshold, the sequence of the second projection converges to the intersection of the diagonal manifold and the stefel manifold; if the first value is projected only to the diagonal manifold, when a second difference is smaller than a second preset threshold, a sequence of second projections converges to the diagonal manifold; if the first value is projected only onto the stille manifold, the sequence of second projections converges on the stille manifold when the second difference is smaller than a second preset threshold;

the access network device may output the update equation.

In the embodiment of the application, a specific scheme for constructing the update equation is provided, so that the power pattern generated by the wide-coverage precoding matrix is close to the target wide-coverage power pattern, and the realizability of the scheme is improved.

Based on the fifteenth implementation manner of the embodiment of the present application, in a sixteenth implementation manner of the embodiment of the present application, after the access network device determines whether a second difference between a result of a current second projection and a result of a previous second projection is smaller than a second preset threshold, before the access network device sums up products of the precoding matrix, the search direction, and the step size to obtain a first numerical value, the method further includes:

if the access network device determines that the second difference is greater than or equal to a second preset threshold, the access network device determines that the sequence of the second projection is not converged, and the access network device returns to perform the step of summing the products of the precoding matrix, the search direction and the step size to obtain a first numerical value.

A second aspect of an embodiment of the present application provides an access network device, where the access network device has a function of implementing an access network device behavior in the first aspect. The function can be realized by hardware, and can also be realized by executing corresponding software by hardware. The hardware or software includes one or more modules corresponding to the functions described above.

A third aspect of the embodiments of the present application provides an access network device, including:

a memory for storing a program;

a processor for executing the program stored in the memory, the processor being configured to perform any one of the possible implementations of the above aspects or aspects when the program is executed.

Alternatively, the memory may be a physically separate unit or may be integrated with the processor.

In one implementation of the third aspect, the access network device may be a chip.

A fourth aspect of embodiments of the present application provides a computer storage medium for storing computer software instructions for an access network device of the second and third aspects described above, including a program for executing the program designed for the access network device.

A fifth aspect of embodiments of the present application provides a computer program product, where the computer program product includes computer software instructions, and the computer software instructions may be loaded by a processor to implement the method flow in the first aspect.

Drawings

FIG. 1A is a diagram of a massive MIMO system architecture according to an embodiment of the present application;

fig. 1B is a schematic diagram of an embodiment of a massive MIMO precoding transmission method in an embodiment of the present application;

FIG. 2 is a schematic diagram illustrating a comparison of results of a wide coverage precoding transmission method and a precoding transmission method based on a ZC sequence when a large-scale MIMO system employs omni-directional transmitting antennas in an embodiment of the present application;

fig. 3 is a diagram illustrating a result of a wide coverage precoding transmission method when a large-scale MIMO system employs directional transmitting antennas in an embodiment of the present application;

FIG. 4 is a flowchart illustrating the setting of search direction using the steepest descent method according to an embodiment of the present disclosure;

FIG. 5 is another flowchart of setting the search direction using the steepest descent method in the embodiment of the present application;

FIG. 6 is another flowchart of setting the search direction using the steepest descent method in the embodiment of the present application;

FIG. 7 is a flowchart illustrating an embodiment of updating a precoding matrix using a steepest descent method;

FIG. 8 is another flowchart of updating a precoding matrix using a steepest descent method in an embodiment of the present application;

FIG. 9 is another flowchart of updating a precoding matrix using a steepest descent method in an embodiment of the present application;

FIG. 10 is another flow chart of the method of using the steepest descent in an embodiment of the present application;

fig. 11 is a schematic diagram of an embodiment of an access network device in an embodiment of the present application;

fig. 12 is a schematic diagram of another embodiment of an access network device in the embodiment of the present application.

Detailed Description

The embodiment of the application provides a large-scale MIMO precoding transmission method and access network equipment, which are used for realizing wide-angle coverage of transmitting power when a large-scale MIMO system transmits transmitting signals through each transmitting antenna, so that the utilization efficiency of the transmitting power of the transmitting signals is improved.

The multiple-input multiple-output technology (multiple-input multiple-output) technology is characterized in that a plurality of transmitting antennas and receiving antennas are respectively used at a transmitting end and a receiving end, signals are transmitted and received through the plurality of antennas at the transmitting end and the receiving end, and therefore communication quality is improved. The large-scale MIMO technology is based on a traditional MIMO system, the number of receiving and transmitting antennas can be increased to hundreds of antennas, the antennas are intensively placed in a large-scale array mode and distributed to a plurality of users in a coverage area of a base station, a plurality of messages of a plurality of terminals can be transmitted on the same time and frequency resources, namely on the same time frequency resource, the antennas are communicated with the base station at the same time by utilizing the spatial freedom degree provided by the large-scale antenna configuration of the base station, so that the maximum beam forming gain and the minimum interference are obtained, the multiplexing capability of the frequency spectrum resources among a plurality of users, the frequency spectrum efficiency of each user link and the capability of resisting the inter-cell interference are improved, the integral utilization rate of the frequency spectrum resources is greatly improved, and the theoretical communication capacity can be infinitely large.

The massive MIMO technology has three major workflows in use: channel estimation, precoding and weight calculation. 1. Channel estimation: that is, the receiving end determines the state (uncertainty) of the wireless transmission channel by processing data of the transmitting end. 2. Pre-coding: the transmitting end often contains a precoder that is capable of generating a precoding matrix using the channel state information for performing a preprocessing operation on the transmitted signal. 3. Calculating a weight value: that is, the precoding matrix and the transmit data stream are used to perform matrix multiplication, so as to realize the exchange of K user data streams to M antenna data streams.

Referring to fig. 1A, fig. 1A is a system architecture diagram of massive MIMO. As shown in fig. 1A, the access network device 101' may be an evolved Node B (eNB or e-NodeB) macro base station, a micro base station (also referred to as a "small base station"), a pico base station, an Access Point (AP), a Transmission Point (TP), or a NodeB (new generation base station) in an LTE system, a next generation mobile communication system, or an authorized assisted access long-term evolution (LAA-LTE) system.

The Terminal devices 102 'and 103' may be referred to as a User Equipment (UE), a Mobile Station (MS), a Mobile Terminal (Mobile Terminal), an intelligent Terminal, and the like, and may communicate with one or more core networks through a Radio Access Network (RAN). For example, the terminal equipment may be a mobile phone (or so-called "cellular" phone), a computer with a mobile terminal, etc., and the terminal equipment may also be a portable, pocket, hand-held, computer-included or vehicle-mounted mobile device and terminal equipment in future NR networks, which exchange voice or data with a radio access network. Description of terminal device: in this application, the terminal device may further include a Relay, and the terminal device and the base station that can perform data communication may be regarded as the terminal device.

It should be noted that the system architecture diagram described in this document is only an example and does not represent a limitation to the specific embodiments of the present application.

The technical solutions of the embodiments of the present application are described in detail below with reference to specific embodiments, and it should be understood that the following specific embodiments are only used for illustrating the present invention and are not used to limit the protection scope of the present invention.

The embodiment of the application is mainly suitable for a large-scale MIMO system with a large-scale antenna array arranged on a base station side to serve multiple users simultaneously.

It should be noted that the embodiments of the present application are not only applicable to the specific system models in the following specific embodiments, but also applicable to system models in other configurations.

Referring to fig. 1B, fig. 1B is a schematic diagram of an embodiment of a massive MIMO precoding transmission method in an embodiment of the present application.

An embodiment of a massive MIMO precoding transmission method in an embodiment of the present application includes:

101. the access network equipment determines a target function according to a preset target wide coverage power pattern;

in this embodiment, in order to obtain a wide-angle coverage range of common signal transmission, the access network device may preset a target wide coverage power pattern, where the target wide coverage power pattern is related to a preset angle and power value of a transmission signal, and then determine a target function according to the preset target wide coverage power pattern.

It should be noted that "wide coverage" referred to in this application refers to an angular range included in a direction in which the access network device transmits a transmission signal through the transmission antenna.

First, in order to set the target wide coverage power pattern, system configuration is first required.

In cellular communication systems, sectorized transmission of signals is typically employed for geographic areas. In a massive MIMO broadcasting system, a massive antenna array configured by a base station has a plurality of sectors, and each sector is composed of M antenna units. Each antenna unit may employ an omni-directional antenna or a 120 degree sector antenna. The spacing between adjacent antenna elements is denoted d and the carrier wavelength is denoted lambda. Large-scale antenna arrays (hereinafter collectively referred to as transmit antennas) may also employ circular arrays or other conveniently mounted array structures. Each antenna unit in the large-scale transmitting antenna is connected with the digital baseband processing unit through a respective transceiving radio frequency unit, an analog-digital/digital-analog conversion unit, a digital optical module and an optical fiber transmission channel.

For a uniformity array, the steering vector can be expressed as:

wherein, M represents the number of antennas, e represents the number of directions of a single antenna element, d represents the distance between adjacent antenna elements, and N is an integer greater than 0.

For a system that simultaneously transmits N data streams, the resulting power pattern can be expressed as a function of the precoding matrix W, i.e.:

where e (θ) denotes a pattern of a single transmission antenna. It is assumed here that all elements have the same transmit antenna pattern, W^HAnd representing the conjugate transpose of the wide-coverage precoding matrix, wherein N is an N-dimensional signal generated by the access network equipment, and N is an integer greater than 0.

Assuming that g (θ) is oversampled by a coefficient K, the discrete azimuth angle can be defined as

The discrete azimuth is a uniform sampling of KM points at azimuth theta. The power pattern g (θ) in KM discrete angular directions can thus be expressed as a column vector with KM elements, i.e.:

wherein the content of the first and second substances,

wherein the content of the first and second substances,

represents the steering vector, | e (θ) | represents the directional pattern of the transmitting antenna, and d represents the spacing of adjacent antenna elements.

Referring to fig. 2 and fig. 3, which will be described later, fig. 2 is a schematic diagram illustrating a comparison between the results of the wide coverage precoding transmission method and the results of the precoding transmission method based on the ZC sequence in the embodiment of the present application when the access network device employs an omni-directional transmitting antenna; fig. 3 is a schematic diagram of a result of transmitting a transmission signal by using a wide coverage precoding transmission method in the embodiment of the present application when an access network device employs a directional transmission antenna. In both fig. 2 and 3, the physical azimuth angle is indicated by the symbol "∘A target power pattern of θ, let f (θ) denote the target power pattern toward the physical azimuth θ. Similarly, f (θ) is defined as a KM-dimensional vector

The value of each element of (a) is set by a user,

is the power pattern g (theta) in KM discrete angular directions.

Secondly, setting an objective function of wide coverage precoding

To approximate the target power pattern (see fig. 2 and 3), the formula for the objective function can be defined as:

wherein the content of the first and second substances,

the power pattern is overlaid for the target,

the power pattern generated for the wide coverage precoding matrix W,

is composed of

The transposing of (1).

In this case, the objective function is a function representing the cosine distance between the power pattern generated by the wide coverage precoding matrix and the target wide coverage power pattern.

To approximate the target power pattern (fig. 2 and 3), the formula for the target function can also be defined as:

wherein the content of the first and second substances,

is composed of

The square of the transpose of (a),

is prepared by

The method comprises the following steps of (1) making a square,

is composed of

The development of (1).

In this case, the objective function is a function representing the cosine distance of the evolution of the power pattern generated by the precoding matrix from the evolution of the target wide coverage power pattern.

In this embodiment, a design example of one target power pattern is specifically described. A massive MIMO system, already equipped with a uniform linear array, is envisaged in the system configuration, where the spacing of the transmit antennas (i.e. antenna elements) is set to half a wavelength. Assume that the target power pattern is represented by the following raised cosine equation:

wherein epsilon is a roll-off factor. To generate a 120 degree sector shaped power pattern to ensure 120 degree sector coverage, θ is set_cIs composed of

Order to

In this case, the power pattern is mainly concentrated in the interval

According to the definition of the discrete azimuth angles: w^optThe target power pattern f (θ) may be rewritten as a column vector of KM discrete azimuth samples, where K represents the oversampling factor and M represents the number of antennas. The oversampling coefficient K is set to "8" in all the simulation examples described later.

102. The access network equipment processes the pre-coding matrix corresponding to the objective function by using a steepest descent method to obtain a wide coverage pre-coding matrix;

after the access network device determines the target function according to the preset target wide coverage power pattern, the precoding matrix corresponding to the target function may be processed by using the steepest descent method to obtain a wide coverage precoding matrix, so that the power pattern generated by the wide coverage precoding matrix continuously approaches the target wide coverage power pattern, the wide coverage precoding matrix is an mxn-dimensional precoding matrix, M is the number of transmitting antennas, and N is a signal generated by the access network device.

Further, in order to efficiently utilize power efficiency, the wide coverage precoding matrix may satisfy a constraint that the transmission signals have the same power at the antenna elements of each transmission antenna within a predetermined angular range (which may also be understood as a spatial direction), that is, each antenna element is assigned an equal power weight. When each antenna element is assigned an equal power weight, in an actually operating system, the power amplifiers connected to each antenna will carry equal power, which means that the power capacity of each power amplifier only needs to be slightly higher than the carried power, and the system can operate normally. Conversely, if the antennas are assigned power weights with large differences, it means that the power carried on some power amplifiers is much smaller than the power capacity of the power amplifier, and the power is much smaller than the power capacity, which means waste of power efficiency. Therefore, the waste of power capacity due to equal power weights is minimal, i.e. the antennas are assigned equal power weights to ensure high power efficiency, i.e. to maximize the power efficiency of each rf channel and transmit antenna, with a maximum value of 1. That is, the access network device may make the transmitted signals on all antennas (i.e., M-dimension) have equal average power in the process of making the power pattern generated by the wide-coverage precoding matrix continuously approach to the target wide-coverage power pattern, so that the wide-coverage precoding matrix W may satisfy the following constraint condition:

where W is a wide coverage precoding matrix, W^HFor conjugate transpose of wide-coverage precoding matrices, I_MIs an identity matrix with dimension of M rows and M columns "

"is the sign of the hadamard product.

Further, the wide-coverage precoding matrix belongs to a diagonal manifold (oblique), and it can also be understood that the M manifold is a diagonal manifold.

Another constraint condition is that, in order to maximize the reachable traversal rate under the independent same-distribution channel, the access network device may satisfy the wide-coverage precoding matrix with the condition of the semi-unitary matrix, and the wide-coverage precoding matrix W may also satisfy the following constraint condition:

W^HW＝I_N (9)

wherein, I_NIs an identity matrix with N rows and N columns.

Further, the semi-unitary matrix belongs to the stefel manifold (tiefel), and it is also understood that the N manifold is the stefel manifold.

To sum up, in the embodiment of the present application, the scheme for optimizing the wide coverage precoding matrix needs to satisfy the following conditions:

that is, the access network device approaches j (w) to dB ═ 0, and simultaneously has the same transmission power on each antenna, and maximizes the achievable traversal rate under independent co-distributed channels.

The following description is continued with respect to a design example of the target power pattern specifically described in step 101 in the present embodiment.

After setting the oversampling coefficient K to "8" in all the examples of the simulation described later, and referring next to fig. 2, when a large-scale MIMO system employs an omnidirectional antenna element (i.e., a transmission antenna), the wide-coverage precoding transmission method in the present embodiment is compared with the precoding transmission method based on a ZC sequence. The antenna element pattern (i.e. the transmit antenna pattern) produced by the omni-directional antenna element is isotropic, i.e. | e (θ) | 1. In this case, the matrix in the formula (4)

Has the following forms:

wherein the content of the first and second substances,

a tangent space, V, of the wide-coverage precoding matrix with dimension KM rows and M columns^H(theta) denotes the conjugate transpose of the steering vector, V^H(θ₀)，…，V^H(θ_KM-1) Representing a guide vector V^HThe conjugate transposed column vector of (θ).

Substituting equation (11) into equation (5) (equation (5) is also referred to as a power error cost function) and equation (6) (equation (6) is also referred to as an amplitude error cost function) of the objective function, the optimal precoding matrix W can be obtained by the steepest descent method in the embodiment of the present application^opt。

Referring to fig. 2, fig. 2 shows a corresponding power pattern in a range of-90 degrees to 90 degrees in azimuth, where the abscissa is the azimuth angle measured in degrees and the ordinate is the power pattern measured in decibels (dB). The base station side is an omnidirectional antenna, and the number of the antennas can be assumed to be tens.

Deriving the precoding matrix based on the power error cost function (i.e., the objective function of equation 5) yields a power pattern represented by a resulting pattern 1, and deriving the precoding matrix based on the amplitude error cost function (i.e., the objective function of equation 6) yields a power pattern represented by a resulting pattern 2. Also, pattern 1 and pattern 2 are also compared with patterns generated based on the ZC sequence scheme.

As can be seen from fig. 2, the fluctuation of the power pattern 1 generated by the wide coverage precoding matrix and the fluctuation of the power pattern 2 generated by the semi-unitary matrix obtained in the embodiment of the present application in different spatial angles are small and substantially coincide with the target power pattern, while the fluctuation of the power pattern generated by the ZC sequence scheme in different spatial angles is relatively large. This shows that, in the embodiment of the present application, in the process of obtaining the wide coverage precoding matrix by approaching the target power to zero in the access network device, the maximum rate descent method is used to enable the wide coverage precoding matrix to satisfy two constraint conditions, one is that the power of the transmission signal on each transmission antenna of the omnidirectional antenna is the same, and the other is that the wide coverage precoding matrix is a semi-unitary matrix.

The formula for the traversal rate is as follows:

wherein R is_iidIs the traversal rate under independent same distribution channel, h is the channel, h is^HIs a conjugate transpose of the channel and,

is composed of

Is calculated as a mathematical expectation.

In addition, as can be seen from fig. 2, compared with the precoding matrix of the ZC sequence, the power pattern generated by the precoding matrix provided in the embodiment of the present application has less interference to the adjacent sectors. The interference caused by the two adjacent sectors of pattern 1 and pattern 2 at-60 to-90 degrees and 60 to 90 degrees is significantly smaller than the interference caused by the precoding matrix of the ZC sequence at the two adjacent sectors.

Furthermore, as shown in fig. 2, since pattern 2 substantially coincides with the target power pattern, and pattern 1 slightly deviates from the target power pattern at some angles, the sidelobe suppression of power pattern 2 generated by the precoding matrix (i.e., semi-unitary matrix) obtained based on the minimized amplitude error cost function is better than that of power pattern 1 generated by the precoding matrix (i.e., wide coverage precoding matrix) obtained based on the minimized power error cost function. Minimizing the amplitude error cost function produces sidelobes that are 3 to 5dB lower than minimizing the power error cost function.

When the access network adopts the omnidirectional transmitting antenna, the power patterns of the signals transmitted by the wide coverage precoding transmission method and the precoding matrix transmission method of the ZC sequence in the embodiment of the present invention are compared, and when the access network adopts the directional transmitting antenna, the power pattern of the signals transmitted by the wide coverage precoding transmission method in the embodiment of the present invention is described below.

Referring to fig. 3, fig. 3 is a result of the wide coverage precoding transmission method in this embodiment when a large-scale MIMO system employs directional antenna elements (i.e., directional transmitting antennas). It is assumed here that the antenna element pattern produced by the directional antenna element is cosine-shaped, i.e. | e (θ) | ═ cos (θ) |. In this case, the matrix in equation 4

Has the following forms:

substituting the formula (13) into the power error cost function (5) and the amplitude error cost function (6), respectively, to obtain the optimal precoding matrix W^optCan be obtained by the steepest descent method in the embodiment of the application.

Fig. 3 shows the corresponding power pattern for an azimuth angle in the range of-90 degrees to 90 degrees, with the abscissa being the azimuth angle measured in degrees and the ordinate being the power pattern measured in decibels (dB). The number of antennas on the base station side may be tens.

The power pattern generated by the precoding matrix (wide coverage precoding matrix) obtained based on the power error cost function is pattern 1, and the power pattern generated by the precoding matrix (semi-unitary matrix) obtained based on the amplitude error cost function is pattern 2. Similar to the case when the omni-directional antenna array element is used, the sidelobe suppression of the power pattern 2 generated by the precoding matrix (semi-unitary matrix) obtained based on the minimized amplitude error cost function is better than the power pattern 2 generated by the precoding matrix (wide coverage precoding matrix) obtained based on the minimized power error cost function. Minimizing the amplitude error cost function produces sidelobes that are 3 to 5dB lower than minimizing the power error cost function.

In addition, as can be seen from fig. 3, the fluctuation (power pattern 2) generated by minimizing the amplitude error cost function is slightly larger than the fluctuation (power pattern 1) generated by minimizing the power error cost function in the main lobe range.

103. The access network equipment multiplies the N-dimensional signal by a wide coverage pre-coding matrix to obtain an M-dimensional transmitting signal;

the access network equipment processes the pre-coding matrix corresponding to the objective function by using a steepest descent method to obtain an M multiplied by N dimensional wide coverage pre-coding matrix, and then multiplies the N dimensional signal by the wide coverage pre-coding matrix to obtain an M dimensional transmitting signal, wherein the N dimensional signal is generated by the access network equipment intermittently.

104. And the access network equipment sends the M-dimensional transmitting signal.

The access network equipment transmits M-dimensional transmission signals through the transmission antennas (i.e., M antenna elements in the antenna array element).

In the embodiment of the application, the access network equipment can determine the target function according to the preset target wide coverage pattern, and then process the precoding matrix corresponding to the target function by using the steepest descent method to obtain the wide coverage precoding matrix, so that the power pattern generated by the wide coverage precoding matrix approaches to the target wide coverage power pattern, and therefore, the wide-angle coverage of the transmitting power in the transmission of the transmitting signals on each transmitting antenna by the large-scale MIMO system can be realized, and the utilization efficiency of the transmitting power is improved.

Further, in order to improve the power efficiency of the transmitted signal, the wide-coverage precoding matrix can satisfy two constraints by the steepest descent method mentioned in step 102. The first is to make the signal transmitting power on each transmitting antenna equal power in the process of making the objective function approach to zero by the access network device, thereby maximizing the power efficiency of each radio frequency channel and transmitting antenna, and the second is to make the wide coverage precoding matrix a semi-unitary matrix, thereby maximizing the reachable traversal rate under independent same distribution channels. The steepest descent method in the examples of the present application is described in detail below.

First, gradient and projection on manifold

In the embodiment of the present application, the manifold may be used to perform an optimal design on the wide coverage precoding matrix, and this section introduces how to calculate the gradient and projection on the manifold.

Referring to fig. 4, fig. 4 is a flowchart illustrating a method for setting a search direction using the steepest descent method according to an embodiment of the present disclosure. Order to

Represents the euclidean gradient of the cost function j (W) with respect to W. The euclidean gradient can be calculated by the following formula:

wherein the content of the first and second substances,

covering the power pattern for the target with a power pattern,

a power pattern generated for the wide coverage precoding matrix W.

The euclidean gradient of the cost function is thus:

or

The constraints (8) and (9) correspond to a diagonal manifold (oblique) and a stilfel manifold (tiefel), respectively. The diagonal manifold and the stilfel manifold are denoted as M and N, respectively. The gradient of j (w) over the manifold M can be calculated according to equation (15) or (16):

wherein the content of the first and second substances,

the gradient of J (W) over manifold N can also be calculated:

wherein the content of the first and second substances,

let ε represent the Euclidean space, then a point in Euclidean space is assembled in X direction

The projection of (a) is defined as:

further has a W-directional manifold

Projection of

And a W-direction manifold

Projection of

The projection may map the vector in the tangent space to the manifold itself. The gradient on the manifold can be viewed as the projection of the euclidean gradient at point W into the tangent space. For example according to the following equations (22) and (23):

in the above, how to perform gradient and projection on the manifold is described, and the precoding design in the embodiment of the present application is described in detail below based on the gradient and projection on the manifold.

Second, precoding design

This section introduces the design algorithm of wide coverage precoding over the intersection of two manifolds.

2.1 problem statement

In the embodiment of the application, the equal power constraint condition and the semi-unitary constraint condition are regarded as two manifolds. Therefore, the access network device can make the wide coverage precoding matrix meet the condition that the power of the transmitted signal on each transmitting antenna is equal power while making the objective function approach to zero, can also make the wide coverage precoding matrix become a semi-unitary matrix, or can make the wide coverage precoding matrix meet the two constraint conditions at the same time.

In the embodiment of the present application, the most preferable scheme is to make the wide-coverage precoding matrix satisfy two constraints at the same time, in which case equation (10) can be rewritten as:

in order to make the wide-coverage precoding matrix satisfy the above formula (24) as a constraint condition, the embodiment of the present application may extend the steepest descent method on a single manifold to the steepest descent method on the intersection of two manifolds.

2.2 Algorithm for setting search Direction

It should be noted that the search direction in the embodiment of the present application is a direction in which the search objective function approaches zero.

The gradient over the intersection of the two manifolds is defined as:

where k is the number of cycles for setting the search direction and updating the precoding matrix (this scheme will be described in detail later).

Then, the steepest descent method is adopted to optimize at the intersection of the two manifolds, and the required search direction is:

to calculate

The embodiment of the present application adopts the alternative projection method proposed by von neumann, i.e., approximates the result by using the following formula (even if the objective function approaches zero):

order to

The initial gradient is indicated.

According to the alternative projection method, the image is projected,

the objective function can be made to approach zero by the following update equation:

where t is the number of cycles to set the search direction.

According to the above analysis, referring to fig. 4, in the steepest descent method of fig. 4, setting the search direction includes the following steps:

401. calculating Euclidean gradient of target function by access network equipment

402. Access network device euclidean gradient

Alternately projecting the cutting space of the syncline manifold (M manifold) and the cutting space of the Stefel manifold (N manifold);

403. the access network equipment judges whether a first difference value between a current first alternate projection result and a previous first alternate projection result is smaller than a preset threshold value or not;

if the first difference is smaller than the preset threshold, go to step 404; otherwise, the access network device determines that the sequence of the first alternative projections is not converged, and returns to execute step 401.

404. The access network device determines that the sequence of first alternating projections converges to a gradient at the intersection of the oblique manifold (M) and the stefel manifold (N).

When the access network device determines that the first difference is less than the preset threshold, the access network device determines that the sequence of the first alternating projections converges to a gradient on an intersection of the oblique manifold (M) and the stetifer manifold (N). The search direction is the inverse of the gradient at the intersection of M and N.

It should be noted that in the example of setting the search direction described in fig. 4, the steepest descent method makes the wide coverage precoding matrix satisfy the above two constraints at the same time, so that it is necessary to set the M manifold as the diagonal manifold and the N manifold as the stille manifold. However, in the embodiment of the present application, the wide-coverage precoding matrix may also satisfy only one constraint condition of the above two constraint conditions.

Referring to fig. 5, fig. 5 is another flowchart of setting a search direction in the steepest descent method. In the following example, the wide coverage precoding matrix only satisfies a constraint condition, that is, the transmitted signals on each transmitting antenna are equal power in the process of making the objective function approach to zero. In the steepest descent method of fig. 5, setting the search direction includes the steps of:

501. calculating Euclidean gradient of target function by access network equipment

502. Access network device euclidean gradient

Projecting to the oblique flow-shaped cutting space;

503. the access network equipment judges whether a first difference value between a current first projection result and a previous first projection result is smaller than a preset threshold value or not;

if the first difference is smaller than the preset threshold, executing step 504; otherwise, the access network device determines that the sequence of the first projection is not converged, and returns to execute step 501.

504. The access network device determines that the sequence of first projections converges to a gradient of the skewed manifold.

And when the access network equipment determines that the first difference is smaller than the preset threshold, the access network equipment judges that the sequence of the first projection converges to the gradient of the oblique manifold. The search direction is the inverse of the gradient of M.

Referring to fig. 6, fig. 6 is another flowchart of setting a search direction in the steepest descent method. In the following example, the wide-coverage precoding matrix only satisfies a constraint condition, that is, the wide-coverage precoding matrix is a semi-unitary matrix in the process of making the objective function approach to zero. In the steepest descent method of fig. 6, setting the search direction includes the steps of:

601. calculating Euclidean gradient of target function by access network equipment

602. Access network device euclidean gradient

Projecting to a cut space of the Stefel manifold;

603. the access network equipment judges whether a first difference value between a current first projection result and a previous first projection result is smaller than a preset threshold value or not;

if the first difference is smaller than the preset threshold, go to step 604; otherwise, the access network device determines that the sequence of the first projection is not converged, and returns to execute step 601.

604. The access network device determines that the sequence of first projections converges to a gradient of the stille manifold.

When the access network device determines that the first difference is smaller than the preset threshold, the access network device determines that the sequence of the first projection converges on the gradient of the stille manifold. The search direction is the inverse of the gradient of M.

It should be noted that, in the examples of fig. 4 to 6, the process of setting the search direction may be executed repeatedly until the search direction is determined.

2.3 precoding matrix update Algorithm

Obtaining search directions using the algorithms provided in the previous section

The optimal point of the objective function defined on the intersection of the manifolds M and N can then be searched using the search direction in the most preferred scheme of fig. 5 above. However, when the search is along direction D^(k)Go on, point W^(k)+α^(k)D^(k)Will leave

And

the intersection of (a). For the steepest descent method on the intersection of two manifolds, W is^(k)+α^(k)D^(k)From

Mapping to

Is necessary.

To ensure that a point can return to the intersection of the two manifolds, the concept of projecting to the intersection of the two manifolds can be introduced in the update equation. By adopting an alternative projection method of projecting to two manifolds respectively, the value of projecting to the intersection of the two manifolds can be approximated, namely:

whereinα is the step size, α^(k)D^(k)K is the number of cycles of the procedure of setting the search direction and the procedure of updating the precoding matrix, W^(k)+α^(k)D^(k)The result of summing the precoding matrix with the product of the search direction and the step size.

Projection (projector)

Can combine W^(k)+α^(k)D^(k)Projection reflux shape

And

the intersection of (a).

Order to

Indicating the initial point. According to the alternative projection method, the image is projected,

can be approximated by the following update equation:

wherein τ is the cycle number of updating the precoding matrix.

Referring to fig. 7, fig. 7 is a flowchart of updating a precoding matrix in the steepest descent method. In the steepest descent method in the example of fig. 7, the updating algorithm of the precoding matrix includes the following steps:

701. access network equipment will cover precoding matrix W a lot^(k)And search direction D^(k)And step size alpha^(k)Summing the products of (a);

702. the access network equipment performs alternate projection on the result obtained by summing in the step 701 towards an oblique manifold (M) and a Stefel manifold (N);

703. the access network equipment judges whether a second difference value between a current second alternate projection result and a previous second alternate projection result is smaller than a preset threshold value or not;

if the second difference is smaller than the preset threshold, go to step 704; otherwise, the access network device determines that the second alternative projection sequence does not converge, and returns to execute step 701.

704. The access network device determines that the sequence of second alternating projections converges to the intersection of the oblique manifold and the stefel manifold.

When the access network device determines that the second difference is smaller than the preset threshold, the access network device determines that the sequence of the second alternating projection converges on a gradient on an intersection of the oblique manifold (M) and the stetifer manifold (N). Convergent

Is the output result W of the required update equation^(k+1)。

It should be noted that, in the example of updating the precoding matrix described in fig. 7, the steepest descent method makes the wide coverage precoding matrix satisfy the above two constraints at the same time, so that it is necessary to set the M manifold as the oblique manifold and the N manifold as the stille manifold. However, in the embodiment of the present application, the wide-coverage precoding matrix may also satisfy only one constraint condition of the above two constraint conditions.

Referring to fig. 8, fig. 8 is another flowchart for setting a search direction in the steepest descent method. In the following example, the wide coverage precoding matrix only satisfies a constraint condition, that is, the transmitted signals on each transmitting antenna are equal power in the process of making the objective function approach to zero. In the steepest descent method of fig. 8, setting the search direction includes the steps of:

801. access network equipment will cover precoding matrix W a lot^(k)And search direction D^(k)And step size alpha^(k)Summing the products of (a);

802. the access network equipment performs alternate projection on the result obtained by summing in the step 801 towards an oblique manifold and a Stefel manifold;

803. the access network equipment judges whether a second difference value between a current second alternate projection result and a previous second alternate projection result is smaller than a preset threshold value or not;

if the second difference is smaller than the preset threshold, executing step 804; otherwise, the access network device determines that the second alternating projected sequence does not converge and returns to perform step 801.

804. The access network device determines that the sequence of second projections converges to a skewed manifold.

And when the access network equipment determines that the second difference is smaller than the preset threshold, the access network equipment judges that the sequence of the second projection converges to the gradient of the oblique manifold. Convergent

Is the output result W of the required update equation^(k+1)。

Referring to fig. 9, fig. 9 is another flowchart for setting a search direction in the steepest descent method. In the following example, the wide-coverage precoding matrix only satisfies a constraint condition, that is, the wide-coverage precoding matrix is a semi-unitary matrix in the process of making the objective function approach to zero. In the steepest descent method of fig. 9, the updating algorithm of the precoding matrix includes the following steps:

901. access network equipment will cover precoding matrix W a lot^(k)And search direction D^(k)And step size alpha^(k)Summing the products of (a);

902. the access network device projects the result obtained by summing in step 901 to the stille manifold;

903. the access network equipment judges whether a second difference value between a current second projection result and a previous second projection result is smaller than a preset threshold value or not;

if the second difference is smaller than the preset threshold, executing step 904; otherwise, the access network device determines that the sequence of the second projection is not converged, and returns to execute step 901.

904. The access network device determines that the sequence of second projections converges to a stilt manifold.

When the access network equipment determines that the second difference value is smaller than the preset threshold value, the access network equipment judges that the second time is putThe sequence of shadows converges to the gradient of the stilfel manifold. Convergent

Is the output result W of the required update equation^(k+1)。

It should be noted that, in the embodiments in fig. 7 to fig. 9, the process of updating the precoding matrix may be performed in a plurality of cycles until the precoding matrix is updated.

2.4 steepest descent method integrating the search direction algorithm and the precoding matrix update algorithm

By using the method of alternating projection to the two manifolds, respectively, the wide coverage precoding matrix can be projected to the intersection of the two manifolds. In addition, the desired search direction is then defined by the direction of the negative gradient at the intersection of the two manifolds.

Referring to fig. 10, fig. 10 is another flowchart of the method using the steepest descent in the embodiment of the present application.

In summary, in the embodiment of the present application, the steepest descent method that combines the algorithm of the search direction and the update algorithm of the wide coverage precoding matrix into a whole includes the following steps:

1001. the access network equipment sets the search direction D according to the gradient on the intersection of the oblique manifold (M) and the Stefel manifold (N)^(k)；

In this embodiment, this step 1001 may be a summary of steps 401 to 404 in the embodiment of fig. 4, and similarly, this step 1001 may also be a summary of steps 501 to 504 in the embodiment of fig. 5, and may also be a summary of steps 601 to 604 in the embodiment of fig. 6, which are not described again here.

1002. The access network equipment sets the step length alpha according to the armijo linear search^(k)；

In this embodiment, the access network device may set the step size according to the armijo linear search, or may set the step size manually and directly, which is not limited herein.

1003. The access network equipment constructs an updating equation by utilizing the searching direction and the step length to update the wide coverage precoding matrix W^(k+1)；

Referring to the above equation (29), it can be seen that:

the access network equipment utilizes the search direction and the step size to construct an update equation, namely, a process of summing the precoding matrix and the product of the search direction and the step size. As previously mentioned, where α is the step size, α^(k)D^(k)K is the number of cycles of the procedure of setting the search direction and the procedure of updating the precoding matrix, W^(k)+α^(k)D^(k)The result of summing the precoding matrix with the product of the search direction and the step size.

1004. The access network equipment respectively carries out alternate projection to the oblique manifold and the Stefel manifold according to the result obtained by the updating equation;

the access network equipment sums the products of the precoding matrix and the search direction and the step length by using an update equation established by the search direction and the step length, and then alternately projects the result obtained by the summation to the oblique manifold and the Stefel manifold respectively.

It should be noted that, in the embodiment of the present application, the result obtained by summing may be projected only to the diagonal manifold, or the result obtained by summing may be projected only to the stills manifold, that is, the steepest descent method in the embodiment of the present application may also enable the wide coverage precoding matrix to only satisfy one constraint condition of the two constraint conditions, which is not limited herein.

1005. The access network equipment judges whether the difference value between the current wide-coverage pre-coding matrix and the last wide-coverage pre-coding matrix is smaller than a preset threshold value or not;

if the difference is smaller than the preset threshold, go to step 1006; otherwise, the access network device determines that the wide coverage precoding matrix is not converged, and returns to execute step 1001.

1006. The access network device determines that the wide-coverage precoding matrix converges to the target precoding matrix.

If the access network equipment determines that the difference value between the current wide-coverage precoding matrix and the previous round of wide-coverage precoding matrix is smaller than the preset threshold value, the access network equipment judges that the wide-coverage precoding matrix converges to the target precoding matrix.

It should be noted that, in the embodiment of the present application, the number of cycles of executing the procedure of setting the search direction may be the same as or different from the number of cycles of executing the procedure of updating the precoding matrix. In addition, in the embodiment of the present application, the cycle number of the process of executing the process of integrating the process of searching the direction and the process of updating the precoding matrix may be the same as or different from the cycle number of the process of executing the process of setting the searching direction; the number of cycles of the procedure for updating the precoding matrix may be the same or different.

The foregoing describes a massive MIMO precoding transmission method in the embodiment of the present application, and the following describes an access network device in the embodiment of the present application, please refer to fig. 11, where fig. 11 is an embodiment of an access network device in the embodiment of the present application.

An embodiment of an access network device in an embodiment of the present application includes:

a first determining unit 1101 configured to determine an objective function according to a preset target wide coverage power pattern;

a processing unit 1102, configured to process a precoding matrix corresponding to the objective function by using a steepest descent method to obtain a wide coverage precoding matrix, where a power pattern generated by the wide coverage precoding matrix approaches a target wide coverage power pattern, the wide coverage precoding matrix is an mxn-dimensional precoding matrix, and M and N are integers greater than 0;

a first calculating unit 1103, configured to multiply the N-dimensional signal by the wide coverage precoding matrix to obtain an M-dimensional transmit signal, where the N-dimensional signal is generated by an access network device;

a sending unit 1104, configured to send the M-dimensional transmission signal, where the wide-coverage precoding matrix is a precoding matrix of wide-angle coverage of transmission power of the transmission signal.

In this embodiment, the processing unit 1102 is specifically configured to make the powers of the M-dimensional transmission signals transmitted on the respective transmission antennas the same.

In this embodiment, the processing unit 1102 is specifically configured to send M-dimensional transmission signals on each sending antenna with the same power, and it needs to satisfy a formula one:

the first formula is:

"is the sign of the hadamard product.

In this embodiment, the wide coverage precoding matrix belongs to a diagonal manifold.

In this embodiment, the processing unit 1102 is specifically configured to, when the access network device satisfies formula two, use the wide coverage precoding matrix as a semi-unitary matrix:

the second formula is: w^HW＝I_N，

Wherein, I_NIs an identity matrix with N rows and N columns.

In this embodiment, the semi-unitary matrix belongs to the stilfel manifold.

In this embodiment, the objective function is a function representing a cosine distance between a power pattern generated by the wide coverage precoding matrix and a target wide coverage power pattern, and the formula of the objective function is the following formula three:

the third formula is:

wherein the content of the first and second substances,

covering the power pattern for the target with a power pattern,

a power pattern generated for the wide coverage precoding matrix W,

is composed of

The transposing of (1).

In this embodiment, the objective function is a function representing a cosine distance between an evolution of a power pattern generated by the precoding matrix and an evolution of a target wide coverage power pattern, and the formula of the objective function is the following formula four:

the fourth formula is:

wherein the content of the first and second substances,

is composed of

The square of the transpose of (a),

is prepared by

The method comprises the following steps of (1) making a square,

is composed of

The development of (1).

In this embodiment, the target wide-coverage power pattern is a KM-dimensional vector in which values of elements in a wide-coverage precoding matrix are preset, where K is an oversampling coefficient, the power pattern generated by the wide-coverage precoding matrix is a KM-dimensional vector, and the power pattern generated by the wide-coverage precoding matrix is a function of the precoding matrix and the transformation matrix.

In this embodiment, the access network device further includes:

a second calculating unit 1105, configured to generate a KM-dimensional vector according to formula five:

the fifth formula is:

wherein the content of the first and second substances,

a tangent space of a wide coverage precoding matrix with dimension KM rows and M columns,

the conjugate transpose of the tangent space of the wide-coverage precoding matrix whose dimension is KM rows and M columns.

In this embodiment, the transformation matrix is a function of the steering vector and the directional pattern of the transmitting antenna of the access network device, and the formula of the transformation matrix is the following formula six:

the formula six is:

wherein the content of the first and second substances,

In this embodiment, the access network device further includes:

a first setting unit 1106, configured to set a search direction according to a gradient on the oblique manifold or the stefel manifold, or according to a gradient on an intersection of the oblique manifold and the stefel manifold, where the search direction is an opposite number of the gradient on the oblique manifold or the stefel manifold, or an opposite number of the gradient on the intersection of the oblique manifold and the stefel manifold, and the search direction is a direction in which a search target function approaches zero;

a second setting unit 1107 for setting a step length;

a constructing unit 1108, configured to construct an update equation using the search direction and the step length, where the update equation is used to update the precoding matrix to an updated precoding matrix;

a first determining unit 1109, configured to determine that the updated precoding matrix converges to the target precoding matrix if a matrix difference between the updated precoding matrix and the precoding matrix is smaller than a preset matrix threshold.

In this embodiment, the access network device further includes:

the second determining unit 1110 is configured to determine that the updated precoding matrix is not converged to the precoding matrix if a matrix difference between the updated precoding matrix and the precoding matrix is greater than or equal to a preset matrix threshold.

In this case, the operations described above are executed again by the first setting unit 1106 to the first determination unit 1109.

In this embodiment, the first setting unit 1106 is specifically configured to calculate a euclidean gradient of the objective function; alternately projecting the Euclidean gradient of the target function to the tangent space of the oblique manifold and the tangent space of the Stefel manifold, or projecting to the tangent space of the oblique manifold, or projecting to the tangent space of the Stefel manifold; if a first difference between a result of the current first projection and a result of a previous first projection is smaller than a first preset threshold, it is determined that the sequence of first projections converges to a gradient at an intersection of the oblique manifold and the stewart manifold, or a gradient of the oblique manifold, or a gradient of the stewart manifold.

In this embodiment, the access network device further includes:

a third determining unit 1111, configured to determine that the sequence of the first projection does not converge on a gradient at an intersection of the oblique manifold and the stetifer manifold, or a gradient of the oblique manifold, or a gradient of the stetifer manifold, if a first difference between a result of the current first projection and a result of the previous first projection is greater than or equal to a first preset threshold.

In this embodiment, the constructing unit 1108 is specifically configured to sum the products of the precoding matrix and the search direction and the step length to obtain a first numerical value; alternately projecting the first numerical value towards an oblique manifold and a Stefel manifold, or projecting towards the oblique manifold, or projecting towards the Stefel manifold; if a second difference between a result of the current second projection and a result of a previous second projection is smaller than a second preset threshold, determining that the sequence of the second projection converges to an intersection of an oblique manifold and a Steftir manifold, or the oblique manifold, or the Steftir manifold; and outputting an updating equation.

In this embodiment, the access network device further includes:

a fourth determining unit 1112, configured to determine that the sequence of the second projection does not converge to an intersection of the oblique manifold and the stetifer manifold, or the oblique manifold, or the stetifer manifold, if a second difference between a result of the current second projection and a result of the previous second projection is greater than or equal to a second preset threshold.

In this case, the construction unit 1108 starts with the step of summing the products of the precoding matrix and the search direction and step size to obtain the first value.

In this embodiment, since the first determining unit 1101 may determine the target function according to a preset target wide coverage power pattern, then the processing unit 1102 uses the steepest descent method to process the precoding matrix corresponding to the target function to obtain a wide coverage precoding matrix, and in the process of using the steepest descent method to make the target function approach to zero, the wide coverage precoding matrix can satisfy the constraint condition of wide angle coverage of the transmit power when transmitting the transmit signal on each transmit antenna, and the wide coverage precoding matrix has the characteristic of a semi-unitary matrix to maximize the reachable traversal rate of each independent distributable channel, so as to finally make the power pattern generated by the wide coverage precoding matrix approach to the target wide coverage power pattern, then multiply the wide coverage precoding matrix to the N-dimensional signal to obtain the M-dimensional transmit signal, and finally the transmitting unit 1104 transmits the M-dimensional transmit signal, therefore, the utilization efficiency of the transmission power of the transmission signal can be improved.

Referring to fig. 12, fig. 12 is a diagram illustrating another embodiment of an access network apparatus according to an embodiment of the present application.

Another embodiment of the access network device in the embodiment of the present application includes:

the access network apparatus 1200 may have a relatively large difference due to different configurations or performances, and may include one or more Central Processing Units (CPUs) 1201 (e.g., one or more processors) and a memory 1205, where one or more applications or data are stored in the memory 1205.

The memory 1205 may be volatile memory or persistent storage, among others. The program stored in the memory 1205 may include one or more modules, each of which may include a sequence of instructions operating on a server. Still further, the central processor 1201 may be configured to communicate with the memory 1205, to execute a sequence of instructional operations on the access network device 1200 in the memory 1205.

The access network apparatus 1200 may also include one or more power supplies 1202, one or more wired or wireless network interfaces 1203, one or more input-output interfaces 1204, and/or one or more operating systems such as Windows Server, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, and the like.

The process executed by the central processing unit 1201 in the access network device 1200 in this embodiment is similar to the method process described in the embodiments shown in fig. 1B and fig. 2 to fig. 10, and is not described again here.

The embodiment of the application has the advantages that after the central processing unit 1201 determines the target function according to the preset target wide coverage power pattern, processing the pre-coding matrix corresponding to the objective function by using a steepest descent method to obtain a wide-coverage pre-coding matrix, and in the process of enabling the objective function to approach zero by using the steepest descent method, the wide-coverage precoding matrix meets the wide-angle coverage of the transmitting power when transmitting the transmitting signal on each transmitting antenna, and the wide-coverage precoding matrix has the characteristic of a semi-unitary matrix, the reachable traversal rate of each independent distributable channel is maximized, thereby finally making the power pattern generated by the wide coverage precoding matrix approach the target wide coverage power pattern, and then, the N-dimensional signal is multiplied by the wide coverage precoding matrix to obtain an M-dimensional transmitting signal, and finally, the M-dimensional transmitting signal is sent by the input/output interface 1204, so that the utilization efficiency of the transmitting power of the transmitting signal can be improved.

Embodiments of the present application also provide a computer storage medium for storing computer software instructions for the foregoing access network device, which includes a program designed for executing the access network device.

Embodiments of the present application also provide a computer program product, which includes computer software instructions that can be loaded by a processor to implement the method flows in the embodiments shown in fig. 1B and fig. 2 to fig. 10.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be substantially implemented or contributed to by the prior art, or all or part of the technical solution may be embodied in a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

Claims

1. A large-scale MIMO precoding transmission method is characterized by comprising the following steps:

the access network equipment determines a target function according to a preset target wide coverage power pattern;

the access network equipment processes the precoding matrix corresponding to the target function by using a steepest descent method to obtain a wide coverage precoding matrix, wherein a power pattern generated by the wide coverage precoding matrix approaches to the target wide coverage power pattern, the wide coverage precoding matrix is an M multiplied by N dimension precoding matrix, and M and N are integers larger than 0;

the access network equipment multiplies the N-dimensional signal by the wide coverage pre-coding matrix to obtain an M-dimensional transmitting signal, wherein the N-dimensional signal is generated by the access network equipment;

the access network equipment sends the M-dimensional transmitting signals on each transmitting antenna, wherein the power of the M-dimensional transmitting signals sent on each transmitting antenna is the same, and the wide-coverage precoding matrix is a precoding matrix covering the transmitting power of the transmitting signals in a wide angle.

2. The method of claim 1, wherein the M-dimensional transmission signals transmitted on the respective transmit antennas have the same power, and satisfy formula one:

the first formula is as follows:

wherein W is the wide coverage precoding matrix, W^HFor the conjugate transpose of the wide-coverage precoding matrix, I_MIs an identity matrix with dimension of M rows and M columns,

is the sign of the hadamard product.

3. The method of claim 2, wherein the wide-coverage precoding matrix belongs to a diagonal manifold.

4. The method of any of claims 1 to 3, wherein when the access network device satisfies formula two, the wide-coverage precoding matrix is a semi-unitary matrix:

the second formula is: w^HW＝I_N，

Wherein, I_NIs an identity matrix with N rows and N columns, and W is the wide coverage precoding matrix.

5. The method of claim 4, wherein the semi-unitary matrix belongs to a Stefel manifold.

6. The method of claim 1, wherein the objective function is a function representing a cosine distance between a power pattern generated by the wide coverage precoding matrix and the target wide coverage power pattern, and the objective function is expressed by formula three:

the third formula is:

wherein the content of the first and second substances,

for the target wide coverage power pattern,

a power pattern generated for the wide coverage precoding matrix W,

is composed of

The transposing of (1).

7. The method of claim 1, wherein the objective function is a function representing a cosine distance of an evolution of a power pattern generated by the precoding matrix from an evolution of the target wide coverage power pattern, and wherein the objective function is formulated as formula four:

the fourth formula is:

wherein the content of the first and second substances,

is composed of

The square of the transpose of (a),

is prepared by

The method comprises the following steps of (1) making a square,

is composed of

The development of (1).

8. The method according to claim 6 or 7, wherein the target wide-coverage power pattern is a KM-dimensional vector with preset values of elements in the wide-coverage precoding matrix, where K is an oversampling factor, and the power pattern generated by the wide-coverage precoding matrix is a KM-dimensional vector and is a function of the precoding matrix and a transformation matrix.

9. The method of claim 8, further comprising:

the access network equipment generates a KM dimensional vector according to a formula five;

the fifth formula is:

wherein the content of the first and second substances,

is the conjugate transpose of the tangent space of the wide-coverage precoding matrix with dimension KM rows and M columns.

10. The method of claim 8, wherein the transformation matrix is a function of steering vectors and a directional pattern of a transmit antenna of the access network device, and wherein the transformation matrix has a formula of six:

the maleThe sixth formula is:

wherein the content of the first and second substances,

represents the steering vector, | e (θ) | represents the directional pattern of the transmitting antenna, d represents the spacing of adjacent antenna elements,

11. the method according to claim 6 or 7, characterized in that the method further comprises:

the access network equipment sets a search direction according to a gradient on an oblique manifold or a Stefel manifold, or sets a search direction according to a gradient on an intersection of the oblique manifold and the Stefel manifold, wherein the search direction is the opposite number of the gradient on the oblique manifold or the Stefel manifold, or is the opposite number of the gradient on the intersection of the oblique manifold and the Stefel manifold, and the search direction is a direction in which a search target function approaches zero;

the access network equipment sets step length;

if the matrix difference between the updated precoding matrix and the precoding matrix is smaller than a preset matrix threshold, the access network equipment judges that the updated precoding matrix is converged to a target precoding matrix.

12. The method of claim 11, wherein setting a search direction according to a gradient on the ramp manifold or the stetifer manifold, or according to a gradient on an intersection of the ramp manifold and the stetifer manifold, by the access network device comprises:

if a first difference between a result of the current first projection and a result of the previous first projection is smaller than a first preset threshold, the access network device determines that the sequence of the first projection converges on a gradient at an intersection of the oblique manifold and the stewart manifold, or a gradient of the oblique manifold, or a gradient of the stewart manifold.

13. The method of claim 11, wherein the access network device constructing an update equation using the search direction and the step size comprises:

if a second difference between a result of a current second projection and a result of a previous second projection is smaller than a second preset threshold, the access network device determines that the sequence of the second projection converges on an intersection of the diagonal manifold and the stefel manifold, or the diagonal manifold, or the stefel manifold;

the access network device outputs the update equation.

14. An access network device, comprising:

the processor, the memory and the input and output equipment are respectively connected with the bus;

the processor is used for determining a target function according to a preset target wide coverage power pattern; processing the precoding matrix corresponding to the objective function by using a steepest descent method to obtain a wide coverage precoding matrix, wherein a power pattern generated by the wide coverage precoding matrix approaches to the target wide coverage power pattern, the wide coverage precoding matrix is an M multiplied by N dimension precoding matrix, the processor multiplies an N dimension signal by the M multiplied by N dimension precoding matrix to obtain an M dimension transmitting signal, and M and N are integers more than 0; multiplying the N-dimensional signal by the wide coverage pre-coding matrix to obtain an M-dimensional transmitting signal, wherein the N-dimensional signal is generated by the access network equipment; and sending the M-dimensional transmitting signals on each transmitting antenna, wherein the power of the M-dimensional transmitting signals sent on each transmitting antenna is the same, and the wide-coverage precoding matrix is a precoding matrix covering the transmitting power of the transmitting signals in a wide angle.

15. The access network device of claim 14, wherein the M-dimensional transmission signals transmitted on the respective transmitting antennas have the same power, and satisfy formula one:

the first formula is as follows:

is the sign of the hadamard product.

16. The access network device of claim 15, wherein the wide-coverage precoding matrix belongs to a diagonal manifold.

17. The access network device of any of claims 14 to 16, wherein the processor is specifically configured to, when the access network device satisfies formula two, the wide-coverage precoding matrix is a semi-unitary matrix:

the second formula is: w^HW＝I_N，

Wherein, I_NThe dimension of the precoding matrix is N rows and N columns of the identity matrix, and W is the wide coverage precoding matrix.

18. The access network device of claim 17, wherein the semi-unitary matrix belongs to a stetefel manifold.

19. The access network device of claim 14, wherein the objective function represents a function of a cosine distance of a power pattern generated by the wide coverage precoding matrix and the target wide coverage power pattern, and the formula of the objective function is formula three:

the third formula is:

wherein the content of the first and second substances,

for the target wide coverage power pattern,

a power pattern generated for the wide coverage precoding matrix W,

is composed of

The transposing of (1).

20. The access network device of claim 14,

the objective function is a function representing a cosine distance between an evolution of a power pattern generated by the precoding matrix and an evolution of the target wide coverage power pattern, and a formula of the objective function is a formula four:

the fourth formula is:

wherein the content of the first and second substances,

is composed of

The square of the transpose of (a),

is prepared by

The method comprises the following steps of (1) making a square,

is composed of

The development of (1).

21. The access network device of claim 19 or 20, wherein the target wide-coverage power pattern is a KM-dimensional vector with preset values of elements in the wide-coverage precoding matrix, where K is an oversampling coefficient, the power pattern generated by the wide-coverage precoding matrix is a KM-dimensional vector, and the power pattern generated by the wide-coverage precoding matrix is a function of the precoding matrix and a transformation matrix.

22. The access network device of claim 21, wherein the processor is further configured to generate a KM-dimensional vector according to the formula five:

the fifth formula is:

wherein the content of the first and second substances,

23. The access network device of claim 21, wherein the transformation matrix is a function of steering vectors and a directional pattern of a transmit antenna of the access network device, and wherein the transformation matrix has the following formula six:

the sixth formula is:

wherein the content of the first and second substances,

represents the steering vector, | e (θ) | represents the directional pattern of the transmitting antenna,

24. the access network device according to claim 19 or 20, wherein the processor is further configured to set a search direction according to a gradient on a ramp manifold or a stefel manifold, or according to a gradient on an intersection of the ramp manifold and the stefel manifold, the search direction being an opposite number of gradients on the ramp manifold or the stefel manifold, or an opposite number of gradients on an intersection of the ramp manifold and the stefel manifold, the search direction being a direction in which the search target function approaches zero; setting a step length; establishing an updating equation by using the searching direction and the step length, wherein the updating equation is used for updating the precoding matrix into an updated precoding matrix; and if the matrix difference value between the updated precoding matrix and the precoding matrix is smaller than a preset matrix threshold, judging that the updated precoding matrix is converged to a target precoding matrix.

25. The access network device of claim 24, wherein the processor is specifically configured to compute a euclidean gradient of the objective function; projecting the Euclidean gradient of the objective function alternately to the tangent space of the oblique manifold and the tangent space of the Stefel manifold, or projecting to the tangent space of the oblique manifold, or projecting to the tangent space of the Stefel manifold; if a first difference between a result of a current first projection and a result of a previous first projection is smaller than a first preset threshold, determining that the sequence of first projections converges on a gradient at an intersection of an oblique manifold and a Stefel manifold, or a gradient of an oblique manifold, or a gradient of a Stefel manifold.

26. The access network device of claim 24, wherein the processor is specifically configured to sum the precoding matrix with the product of the search direction and the step size to obtain a first value; projecting the first value to the oblique manifold and the Stefel manifold alternately, or projecting to the oblique manifold, or projecting to the Stefel manifold; if a second difference between a result of a current second projection and a result of a previous second projection is smaller than a second preset threshold, determining that the sequence of the second projection converges on an intersection of the diagonal manifold and the stetiform, or the diagonal manifold, or the stetiform; and outputting the updating equation.

27. A computer-readable storage medium comprising instructions that, when executed on a computer, cause the computer to perform the method of any of claims 1 to 13.