Channel estimation method, non-volatile storage medium, and computer device

Abstract

The present invention discloses a channel estimation method, a non-volatile storage medium, and a computer device. The method comprises: obtaining an estimated mean of a posterior distribution of a historical channel in a historical time slot as a historical mean; determining an initial mean of a target manifold in a target time slot based on the historical mean, and determining an initial variance of the target manifold in the target time slot based on the initial mean, wherein the target time slot and the historical time slot are adjacent; determining an initial value of a natural parameter of the target manifold based on the initial mean and the initial variance; iteratively updating the initial value of the natural parameter of the target manifold to obtain a target value of the natural parameter of the target manifold; determining a target mean based on the target value of the natural parameter of the target manifold, and estimating a target channel in the target time slot. The present invention solves the technical problem of high computational complexity of channel estimation caused by the large number of iterations required to achieve better channel estimation performance in the simplified information geometry channel estimation algorithm in the related art.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a channel estimation method, a non-volatile storage medium, and a computer device. Background Art

[0002] Massive Multiple Input Multiple Output (MIMO) is one of the key technologies in fifth-generation mobile communications (5G). It can provide huge capacity gains for communication systems and will also play a key role in future 6G systems. The performance of massive MIMO systems is highly correlated with the quality of channel estimation. Pilot-assisted channel estimation is a common channel estimation method in actual systems. The transmitter periodically sends pilot signals, and the receiver uses the received pilot signals to estimate the channel. After receiving the pilot signals, the task of channel estimation is to obtain a posteriori information of the channel parameters. However, in massive MIMO systems, the dimension of the channel matrix is ​​very large, which poses a huge challenge to the acquisition of a posteriori information. Due to the high computational complexity of the inverse operation of large-dimensional matrices, traditional channel estimation algorithms are computationally unbearable.

[0003] The simplified information geometry (SIG) channel estimation algorithm has the advantage of low complexity. The SIG algorithm calculates a posteriori information of channel parameters within the information geometry framework and transforms this calculation into an iterative process. Therefore, its computational complexity is related to the number of iterations.

[0004] The SIG channel estimation algorithm iteratively updates the auxiliary and target manifold parameters, then performs channel estimation based on the target manifold parameters after iterative convergence. Currently, the SIG algorithm initializes the auxiliary and target manifold parameters with zero values ​​in each time slot, and then iteratively updates the parameters starting from zero. This results in slow parameter convergence (typically requiring dozens or even hundreds of iterations to achieve good channel estimation performance). However, excessive iterations lead to excessive computational complexity (computational complexity is approximately proportional to the number of iterations).

[0005] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0006] Embodiments of the present invention provide a channel estimation method, a non-volatile storage medium, and a computer device to at least solve the technical problem of high computational complexity of channel estimation caused by the large number of iterations required to achieve better channel estimation performance in the simplified information geometry channel estimation algorithm in the related art.

[0007] According to one aspect of an embodiment of the present invention, a channel estimation method is provided, including: obtaining an estimated mean of a posterior distribution of a historical channel in a historical time slot as a historical mean; determining an initial mean of a target manifold in a target time slot based on the historical mean, and determining an initial variance of the target manifold in the target time slot based on the initial mean, wherein the target manifold is used to approximate N nonlinear terms corresponding to the channel estimation, and the target time slot and the historical time slot are adjacent time slots; determining initial values ​​of natural parameters of the target manifold based on the initial mean and the initial variance, wherein the natural parameters of the target manifold characterize the mean and covariance matrix of the distribution in the target manifold; iteratively updating the initial values ​​of the natural parameters of the target manifold to obtain target values ​​of the natural parameters of the target manifold; determining a target mean based on the target values ​​of the natural parameters of the target manifold, and using the target mean as an approximate value of the mean of the posterior distribution of the target channel; and performing channel estimation on the target channel in the target time slot based on the target mean.

[0008] Optionally, based on the historical mean, the initial mean of the target manifold in the target time slot is determined, including: obtaining historical object data communicated using the historical channel in the historical time slot, and target object data communicated using the target channel in the target time slot; processing the historical mean based on the historical object data and the target object data to obtain the initial mean.

[0009] Optionally, the historical mean is processed according to the historical object data and the target object data to obtain the initial mean, including: comparing the historical object data and the target object data to obtain a comparison result; obtaining the initial mean based on the comparison result includes at least one of the following: when the historical object data and the target object data are the same, using the historical mean as the initial mean; when there is newly added object data in the target object data, adding a row corresponding to the newly added object data to the historical mean to obtain the initial mean, wherein the newly added object data is data that does not exist in the historical object data, and the historical mean and the initial mean are represented by vectors; when there is leaving object data in the historical object data, deleting a row corresponding to the leaving object data from the historical mean to obtain the initial mean, wherein the leaving object data does not exist in the target object data, and the historical mean and the initial mean are represented by vectors.

[0010] Optionally, determining the initial variance of the target manifold in the target time slot according to the initial mean includes: obtaining target object data communicated using the target channel in the target time slot; and determining the initial variance of the target manifold according to the target object data and the initial mean.

[0011] Optionally, the initial variance of the target manifold in the target time slot is determined based on the target object data and the initial mean, including: determining the energy matrix of the target channel based on the target object data; determining the estimation error generated by adopting the initial mean; and determining the initial variance based on the energy matrix and the estimation error.

[0012] Optionally, the initial values ​​of the natural parameters of the target manifold are iteratively updated to obtain target values ​​of the natural parameters of the target manifold, including: constructing an auxiliary manifold, wherein the auxiliary manifold is used to approximate N-1 nonlinear terms corresponding to the channel estimation; determining the initial values ​​of the natural parameters of the auxiliary manifold based on the initial values ​​of the natural parameters of the target manifold, wherein the natural parameters of the auxiliary manifold characterize the mean and covariance matrix of the distribution in the auxiliary manifold; projecting the auxiliary manifold onto the target manifold to obtain a projection result; updating the natural parameters of the auxiliary manifold based on the projection result, and updating the natural parameters of the target manifold to obtain intermediate values ​​of the natural parameters of the target manifold; using the above method for obtaining the intermediate values ​​of the natural parameters of the target manifold, iteratively updating the intermediate values ​​of the natural parameters of the target manifold to obtain the target values ​​of the natural parameters of the target manifold.

[0013] Optionally, channel estimation is performed on the target channel in the target time slot according to the target mean, including: determining a channel matrix of the target channel according to the target mean; and performing channel estimation on the target channel according to the channel matrix of the target channel.

[0014] According to another aspect of an embodiment of the present invention, a non-volatile storage medium is provided. The non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute any one of the above-mentioned channel estimation methods.

[0015] According to yet another aspect of an embodiment of the present invention, a computer device is provided. The computer device includes a processor, and the processor is configured to run a program. When the program is run, any one of the above-mentioned channel estimation methods is executed.

[0016] According to yet another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program. When the computer program is executed by a processor, any one of the above-mentioned channel estimation methods is implemented.

[0017] In an embodiment of the present invention, an estimated mean of the posterior distribution of a historical channel in an adjacent previous time slot is used to determine the initial mean of a target manifold in a target time slot, by obtaining the estimated mean of the posterior distribution of a historical channel in the historical time slot as the historical mean; according to the historical mean, the initial mean of the target manifold in the target time slot is determined, and according to the initial mean, the initial variance of the target manifold in the target time slot is determined, wherein the target manifold is used to approximate N nonlinear terms corresponding to the channel estimation, and the target time slot and the historical time slot are adjacent time slots; based on the initial mean and the initial variance, the initial values ​​of the natural parameters of the target manifold are determined, wherein the natural parameters of the target manifold characterize the mean and variance of the distribution in the target manifold. Variance matrix; iteratively updating the initial values ​​of the natural parameters of the target manifold to obtain the target values ​​of the natural parameters of the target manifold; determining the target mean according to the target values ​​of the natural parameters of the target manifold, and taking the target mean as the approximate value of the mean of the posterior distribution of the target channel; performing channel estimation on the target channel within the target time slot according to the target mean, thereby achieving the purpose of reducing the number of iterations from the initial values ​​of the natural parameters of the target manifold to the target values, thereby achieving the technical effect of reducing the computational complexity of channel estimation, and further solving the technical problem of high computational complexity of channel estimation caused by the large number of iterations required for the simplified information geometry channel estimation algorithm in the related art to achieve better channel estimation performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0019] Figure 1 A hardware structure block diagram of a computer terminal for implementing a channel estimation method is shown;

[0020] Figure 2 is a schematic flow chart of a channel estimation method according to an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of the effect of test scenario 1 provided according to an optional embodiment of the present invention;

[0022] Figure 4 is a schematic diagram of iterative convergence of test scenario 1 provided according to an optional embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram of the effect of test scenario 2 provided according to an optional embodiment of the present invention;

[0024] Figure 6 This is a schematic diagram of the effect of test scenario three provided according to an optional embodiment of the present invention;

[0025] Figure 7 is a structural block diagram of a channel estimation device provided according to an optional embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] According to an embodiment of the present invention, an embodiment of a method for channel estimation is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0029] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal for implementing a channel estimation method. Figure 1 As shown, the computer terminal 10 may include one or more (illustrated as 102a, 102b, ..., 102n in the figure) processors (the processor may include but is not limited to a microprocessor MCU or a programmable logic device FPGA and other processing devices), a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0030] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry." The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuitry may be a single, independent processing module, or may be incorporated in whole or in part into any of the other components of the computer terminal 10. As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0031] Memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the channel estimation method in the embodiment of the present invention. The processor executes the software programs and modules stored in memory 104 to execute various functional applications and data processing, thereby implementing the channel estimation method of the aforementioned application. Memory 104 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, memory 104 may further include memory remotely located relative to the processor, and such remote memory may be connected to computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0032] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 10 .

[0033] In the related technology, the previous simplified information geometry algorithm used "zero value" to initialize the parameters of the auxiliary manifold and the target manifold when performing channel estimation in each time slot. That is to say, the parameters of the auxiliary manifold and the target manifold were given the initial value "zero" in each time slot, and iterations started from "zero". Therefore, more iterations are required to obtain better performance, and too many iterations will lead to higher complexity of the algorithm (the complexity of the algorithm is approximately proportional to the number of iterations).

[0034] In the related art, considering the multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) system, the following model can be established for the MIMO system: the base station is equipped with N r =N r,v ×N r,h The uniform array of antennas has N antennas in the vertical and horizontal directions. r,v and N r,h The number of users is K, and each user is equipped with a single antenna. For Orthogonal Frequency Division Multiplexing Modulation (OFDM), the number of subcarriers is N. c , the system sampling interval is T s , the cyclic prefix is ​​N g The subcarrier set used for uplink channel estimation is Where N1 and N2 are the starting and last subcarrier index numbers respectively, N p is the number of subcarriers used for channel estimation.

[0035] At the nth subcarrier, the frequency domain received signal is:

[0036]

[0037] in is the channel of the kth user on the nth subcarrier, and K is the total number of users. k [n] is the frequency domain pilot signal sent by the k-th user on the n-th subcarrier. is Gaussian noise, and in, is the mathematical expectation. H represents the calculation of the conjugate transpose, is the noise power, and I is the unit matrix. The spatial frequency domain channel coefficient matrix of user k is:

[0038]

[0039] remember x k =[x k [N1],...,x k [N2]] T , The superscript T represents the transpose of a matrix or vector, and the received signal can be written as:

[0040]

[0041] Bringing the beam domain channel model into play, we can get:

[0042] G k =VH k F T (4)

[0043] in is the statistical beam domain channel matrix, is the delay domain sampling matrix, in, F τ is the delay domain oversampling factor, [F] m,n is the element in the mth row and nth column of the matrix. is the angle domain sampling matrix, where Kronecker product, is the angular domain sampling matrix in the vertical direction, N v =F v N r,v , F v is the oversampling factor in the vertical direction. is the angle domain sampling matrix in the horizontal direction, N h =F h N r,h , F h is the oversampling factor in the horizontal direction. Substituting formula (4) into the received signal model, we can obtain:

[0044] Y=VH a M+Z (5)

[0045] in

[0046] The beam domain energy matrix is ​​defined as:

[0047]

[0048] in H a The conjugate of , ⊙ represents the multiplication of corresponding matrix elements, i.e., dot product. Due to the sparsity of the channel, the values ​​of many elements in the beam domain energy matrix are close to zero.

[0049] By vectorizing formula (5), we can get:

[0050] y=Ah+z (7)

[0051] Let h be vec(H a ) is the vector after removing the term with zero energy, for The matrix after removing the corresponding columns, where is the Kronecker product, N = N r N p , M is H a The number of terms with non-zero variance. y and z are vectorized by Y and Z respectively. The prior distribution of h follows a complex Gaussian distribution, i.e. Where D = diag(d), d is the vector after removing the elements with value 0 from vec{Ω}. Its posterior probability distribution p(h|y) = p(h; μ, Σ) is still a complex Gaussian distribution The mean of the distribution is μ and the covariance is Σ.

[0052] The following is a brief description of the SIG algorithm:

[0053] The prior distribution of h is Gaussian distribution, that is Where D is the energy matrix of the channel. Its posterior distribution is also Gaussian distribution, and the probability density function is:

[0054]

[0055] where y n is the nth element of y, p i is the probability density function of the i-th prior distribution, p n is the nth conditional probability density function, is the nth row of A, and the superscript H indicates the conjugate transpose of a matrix or vector. is the inner product of complex variables (defined as x and y are vectors. X and Y are matrices). Define the function in The vector formed by the energy matrix is ​​d h =f(0,-D -1 ), the vector consisting of channel sufficient statistics is t h =f(h,I⊙(hh H )). C, C' and is the normalization factor. c n (h) is a nonlinear term, written as:

[0056]

[0057] Next, we define N auxiliary manifolds M n , which has the form:

[0058]

[0059] ψ n =f(θ n ,Θ n ) is called the natural parameter, Θ nis an M×M real diagonal matrix, subscript From the above formula, we can see that each auxiliary manifold only retains one nonlinear term and approximates the remaining N-1 nonlinear terms. It is transformed into the traditional Gaussian distribution probability density function form (the mean is denoted by μ n , the variance is recorded as Σ n ), the relationship between the natural parameters and the mean variance on the auxiliary manifold can be calculated as follows:

[0060]

[0061] The target manifold is defined as follows:

[0062]

[0063] where ψ0=f(θ0,Θ0), Θ0 is an M×M real diagonal matrix. Similarly, transforming it into the traditional Gaussian distribution probability density function (mean is μ0, variance is Σ0), the relationship between the natural parameters and the mean variance on the target manifold is as follows:

[0064] Σ0=(D -1 -Θ0) -1

[0065]

[0066] Right now:

[0067]

[0068] The SIG algorithm process in related technologies is as follows:

[0069] First, assign the initial value to the mean μ initial = 0 and variance Σ initial =D, and the initial values ​​of the natural parameters of the target manifold are obtained according to formula (15).

[0070] Secondly, repeat the following process until the predetermined number of iterations is reached: At the tth iteration, M n Perform m-projection to the target manifold M0 (m-projection minimizes KL divergence), which is applied in the method provided by the present invention to make M n Minimize the KL divergence between and M0, and record the obtained natural parameter as

[0071]

[0072] It means that m is projected onto M0. The projection of m is to minimize the KL divergence. By calculation, we can get:

[0073]

[0074] Finally, calculate the message ( For c n (h)), update the auxiliary and target manifold parameters:

[0075]

[0076] When it is considered that each ψ n When they are equal, we can n Denoted as ψ N , the update of auxiliary manifold parameters is changed to:

[0077]

[0078] At this point, the m projection is further updated to obtain the following formula:

[0079]

[0080]

[0081] Adding the damping factor α, the update of the target manifold parameters is:

[0082]

[0083] The above update process is expressed using natural parameters, namely:

[0084]

[0085] And update the auxiliary manifold parameters according to the updated target manifold function:

[0086]

[0087] Finally, the channel estimation result can be obtained by converting the target manifold parameters into the mean and variance of the posterior distribution based on the last obtained result:

[0088]

[0089] If the initial value of the mean μ is set to 0, the corresponding initial value of the variance Σ is D. That is, when the mean is initialized from 0, the parameters of the target manifold and the auxiliary manifold start iterating from "zero value", which results in a large number of iterations required to achieve better performance, and thus the algorithm complexity is higher.

[0090] To solve the above problems, the present invention provides a channel estimation method. Figure 2 FIG. 1 is a flow chart of a channel estimation method according to an embodiment of the present invention. Figure 2As shown, the method includes the following steps:

[0091] Step S201 : obtaining the mean of the posterior distribution estimation of the historical channel in the historical time slot as the historical mean.

[0092] Step S202: Determine an initial mean of the target manifold in the target time slot based on the historical mean, and determine an initial variance of the target manifold in the target time slot based on the initial mean, wherein the target manifold is used to approximate N nonlinear terms corresponding to the channel estimation, and the target time slot and the historical time slot are adjacent time slots.

[0093] In the above two steps, due to the correlation between channels in adjacent time slots, the parameters required for channel estimation in the next time slot can be initialized based on the channel estimation results of the historical channel in the previous time slot. This can achieve higher channel estimation performance with a smaller number of iterations, thereby reducing computational complexity. Specifically, the mean of the posterior distribution obtained by channel estimation of the historical channel in the previous time slot (historical time slot) can be obtained as the historical mean, and the historical mean is used to initialize the parameters required for target channel estimation in the target time slot.

[0094] As an optional embodiment, the initial mean of the target manifold in the target time slot is determined based on the historical mean, including: obtaining historical object data communicated using the historical channel in the historical time slot, and target object data communicated using the target channel in the target time slot; processing the historical mean based on the historical object data and the target object data to obtain the initial mean.

[0095] Optionally, the historical mean of the historical channel can be used as a basis, but since there is still a gap between the historical channel and the target channel, the historical mean can be adjusted based on the historical object data communicated using the historical channel and the target object data communicated using the target channel in the target time slot to obtain an initial mean, and the initial mean is used to perform channel estimation on the target channel.

[0096] As an optional embodiment, the historical mean is processed according to the historical object data and the target object data to obtain the initial mean, including: comparing the historical object data and the target object data to obtain a comparison result; obtaining the initial mean based on the comparison result includes at least one of the following: when the historical object data and the target object data are the same, using the historical mean as the initial mean; when there is newly added object data in the target object data, adding a row corresponding to the newly added object data to the historical mean to obtain the initial mean, wherein the newly added object data is data that does not exist in the historical object data, and the historical mean and the initial mean are represented by vectors; when there is leaving object data in the historical object data, deleting the row corresponding to the leaving object data from the historical mean to obtain the initial mean, wherein the leaving object data does not exist in the target object data, and the historical mean and the initial mean are represented by vectors.

[0097] Optionally, from the historical time slot to the target time slot, the user using the channel changes from the historical object to the target object. The historical mean can be processed based on the user's change. First, the historical object data and the target object data can be compared to obtain a comparison result, i.e., the user's change. Second, the initial mean can be processed in a classified manner: when the historical object data and the target object data are the same, that is, the user using the channel in the historical time slot has not changed compared to the user using the channel in the target time slot, the historical mean can be used as the initial mean. When the target object data contains newly added object data, that is, when the target time slot has a new user compared to the users in the historical time slot, the row corresponding to the newly added user can be added to the vector representing the historical mean to obtain the vector representing the initial mean. When the historical object data contains departing object data, and the departing object data does not exist in the target object data, that is, when the target time slot has a new user compared to the users in the historical time slot, the row corresponding to the departing user can be deleted from the vector representing the historical mean to obtain the vector representing the initial mean.

[0098] Specifically, the present invention proposes solutions for determining the initial mean and initial variance under three scenarios: 1. There are no changes in the number of users in the system; 2. A new user joins the system; 3. An existing user leaves the system.

[0099] 1. When the number of users in the system remains unchanged, the solution for determining the initial mean is as follows:

[0100] There are K users in the system, and the energy matrix of these K users is Where M is the number of non-zero elements in the energy matrix of the K users. The number of iterations of the algorithm in each time slot is set to T. Under the environment where the energy matrix remains unchanged, the mean of the next time slot is initialized using the mean of the previous time slot after T iterations, that is:

[0101]

[0102] in It is the mean value calculated after T iterations of the previous time slot.

[0103] The initialization of the variance needs to match the initialization of the mean. When the variance and the mean are initialized to match, the convergence of the algorithm can be accelerated. Therefore, the present invention proposes to initialize the variance using the error calculation method, that is, to initialize the variance using the following formula:

[0104]

[0105] The definitions of y and A are shown in formula (7). Formula (30) can be understood as follows: Based on the received signal model formula (7), The calculation is done using μ initial,1 Make an estimated error of the mean and then use the normalized prior variance Considering that the posterior variance is the estimation error, the estimation error is first calculated in Equation (30). The estimation error of the element with large energy is also relatively large, so the estimation error is weighted by the normalized prior variance.

[0106] 2. When a new user joins the system, the solution for determining the initial mean is as follows:

[0107] There are K users in the system, and the energy matrix of these K users is Where M is the number of non-zero elements in the energy matrix of the K users. In a certain time slot, K1 new users join, and the energy matrix of the K1 new users is recorded as is a real diagonal matrix, where M1 is the number of non-zero elements in the energy matrix of K1 new users, then the energy matrix of all users in this time slot is in 0 P×Q is a zero matrix of P×Q dimensions. Accordingly, A in formula (7) becomes The solution for determining the initial mean in this scenario is:

[0108]

[0109] in, is an M1×1 dimensional zero matrix.

[0110] 3. When an original user leaves the system, the solution for determining the initial mean is as follows:

[0111] There are K users in the system, and the energy matrix of these K users is Where M is the number of non-zero elements in the energy matrix of the K users. In a certain time slot, K2 users leave the system, and the statistical information becomes Where M2 is the number of non-zero elements in the energy matrices of the remaining K-K2 users, and D2 is the new energy matrix obtained by removing the energy matrix elements of the K2 users who left. Accordingly, A in the received signal model becomes is the mean of the original K users calculated after T iterations of the previous time slot, is the estimated mean vector of the remaining users after removing the estimated mean of the K2 users who left. The solution for determining the initial mean in this scenario is:

[0112]

[0113] It should be noted that the second and third cases can occur simultaneously. That is, the target object data includes both newly added and departed objects. This means that the target time slot has both newly added and departed users relative to the historical time slots. In this case, the rows and columns corresponding to newly added users can be added to the matrix of historical means, while the rows and columns corresponding to departed users can be deleted to obtain the initial mean. This initial mean can then be used to determine the corresponding initial variance for subsequent calculations.

[0114] As an optional embodiment, determining the initial variance of the target manifold in the target time slot based on the initial mean includes: obtaining target object data communicated using the target channel in the target time slot; and determining the initial variance of the target manifold based on the target object data and the initial mean.

[0115] Optionally, the initialization of the variance needs to match the initialization of the mean. Only when the variance and mean are initialized to match can the convergence of the algorithm be accelerated. Initializing the mean (i.e., determining the initial mean) can be combined with initializing the variance (i.e., determining the initial variance). In other words, the initial variance can be determined based on the initial mean. Specifically, the initial variance can be determined based on the change in users from the historical time slot to the target time slot, as well as the initial mean.

[0116] As an optional embodiment, the initial variance of the target manifold in the target time slot is determined based on the target object data and the initial mean, including: determining the energy matrix of the target channel based on the target object data; determining the estimation error generated by adopting the initial mean; and determining the initial variance based on the energy matrix and the estimation error.

[0117] Optionally, the energy matrix of the target channel can be determined based on the user usage in the target time slot, that is, the target object data, and then the initial variance can be calculated based on the initial mean and the energy matrix. First, the estimation error generated by using the posterior mean of the historical channel in the target time slot as the mean can be calculated based on the initial mean, and then the estimation error can be weighted according to the energy matrix to obtain the initial variance. Specifically, the initial mean can be determined based on the historical mean, and then the initial variance can be determined based on the initial mean. Since the initialization of the variance needs to match the initialization of the mean, the variance initialization is also divided into three scenarios: 1. The number of users in the system remains unchanged; 2. A new user joins the system; 3. An original user leaves the system. Specifically, when the number of users in the system remains unchanged, the initial variance can be determined according to formula (30); when a new user joins the system, the initial variance can be determined according to formula (31); when an original user leaves the system, the initial variance can be determined according to formula (32).

[0118] Step S203 : determining initial values ​​of natural parameters of the target manifold based on the initial mean and initial variance, wherein the natural parameters of the target manifold represent the mean and covariance matrix of the distribution in the target manifold.

[0119] In this step, when determining the initial values ​​of the natural parameters of the target manifold, the mean and variance need to be involved in the calculation at the same time. Among them, the initialization of the variance needs to match the initialization of the mean. When the variance matches the initialization of the mean, the convergence of the algorithm can be accelerated. Based on the initialization of the mean (i.e., determining the initial mean), the initialization of the variance (i.e., determining the initial variance) can be further achieved, and the initial values ​​of the natural parameters of the target manifold can be determined based on the initial mean and the initial variance. Among them, the definition of the target manifold is as shown in formula (29), where the natural parameters are θ0 and θ0, and the relationship between the natural parameters and the mean variance on the target manifold is as shown in formula (30). Therefore, the initial values ​​of the natural parameters of the target manifold can be determined based on the initial mean and the initial variance.

[0120] Specifically, since the present invention proposes a solution for determining the initial mean and initial variance in three scenarios, the initial values ​​of the natural parameters of the target manifold can be determined in the three scenarios respectively.

[0121] 1. When the number of users in the system remains unchanged, the initialization scheme is as follows:

[0122] Then, Equations (29) and (30) are substituted into Equation (31) to initialize the natural parameters of the target manifold, and Equations (25) and (26) are substituted into Equation (25) and (26) to initialize the natural parameters of the auxiliary manifold, namely:

[0123]

[0124] 2. When a new user joins the system, substitute formula (31) into formula (33), formula (25) and formula (26) to obtain the initialization of the target manifold and auxiliary manifold parameters:

[0125]

[0126] 3. When the original user leaves the system, substitute formula (32) into formula (35), formula (25) and formula (26) to obtain the initialization of the target manifold and auxiliary manifold parameters:

[0127]

[0128] Step S204: iteratively update the initial values ​​of the natural parameters of the target manifold to obtain target values ​​of the natural parameters of the target manifold.

[0129] Step S205 : determining a target mean value according to the target values ​​of the natural parameters of the target manifold, and using the target mean value as an approximate value of the mean value of the posterior distribution of the target channel.

[0130] In the above two steps, the method of iteratively updating the initial values ​​of the natural parameters of the target manifold in the relevant technology can be used to obtain the target values ​​of the natural parameters of the target manifold, and the mean of the posterior distribution of the target channel can be determined based on the target values ​​of the natural parameters of the target manifold.

[0131] As an optional embodiment, the initial values ​​of the natural parameters of the target manifold are iteratively updated to obtain target values ​​of the natural parameters of the target manifold, including: constructing an auxiliary manifold, wherein the auxiliary manifold is used to approximate N-1 nonlinear terms corresponding to the channel estimation; determining the initial values ​​of the natural parameters of the auxiliary manifold based on the initial values ​​of the natural parameters of the target manifold, wherein the natural parameters of the auxiliary manifold characterize the mean and covariance matrix of the distribution in the auxiliary manifold; projecting the auxiliary manifold onto the target manifold to obtain a projection result; updating the natural parameters of the auxiliary manifold based on the projection result, and updating the natural parameters of the target manifold to obtain intermediate values ​​of the natural parameters of the target manifold; using the above method for obtaining the intermediate values ​​of the natural parameters of the target manifold, iteratively updating the intermediate values ​​of the natural parameters of the target manifold to obtain the target values ​​of the natural parameters of the target manifold.

[0132] Optionally, an auxiliary manifold can be constructed, and the initial values ​​of the auxiliary manifold can be calculated based on the initial values ​​of the natural parameters of the target manifold and the above equations (25) and (26). Then, the m-projection can be calculated according to equations (20) and (21), the target manifold parameters can be updated according to equations (23) and (24), and the auxiliary manifold parameters can be updated according to the updated target manifold. The number of iterations plus 1 is recorded, and the above steps are repeated to iteratively update the natural parameters of the target manifold until the number of iterations T is reached to obtain the target values ​​of the natural parameters of the target manifold. Finally, the target values ​​of the natural parameters of the target manifold are substituted into equations (27) and (28) to convert them into the mean and variance of the posterior distribution to obtain the target mean.

[0133] Step S206: performing channel estimation on the target channel in the target time slot according to the target mean value.

[0134] As an optional embodiment, channel estimation is performed on the target channel in the target time slot according to the target mean, including: determining a channel matrix of the target channel according to the target mean; and performing channel estimation on the target channel according to the channel matrix of the target channel.

[0135] Optionally, the target mean is obtained, that is, h is obtained, and the channel coefficient matrix G can be determined by applying formula (36) k , obtain the channel matrix of the target channel and realize channel estimation of the target channel.

[0136] Through the above steps, the estimated mean of the posterior distribution of the historical channel of the adjacent previous time slot is used to determine the initial mean of the target manifold in the target time slot, thereby achieving the purpose of reducing the number of iterations from the initial value of the natural parameter of the target manifold to the target value, thereby achieving the technical effect of reducing the computational complexity of channel estimation, and further solving the technical problem of high computational complexity of channel estimation caused by the large number of iterations required to achieve better channel estimation performance in the simplified information geometry channel estimation algorithm in the related technology.

[0137] In summary, the process of the improved SIG algorithm of the present invention is summarized as follows:

[0138] 1. Based on the historical mean in the historical time slot, determine the initial mean μ in the target time slot initial and initial variance Σ initial , according to μ initial and Σ initial The initial values ​​of the target manifold parameters are calculated as

[0139] 2. Calculate the m-projection according to formula (20) and formula (21).

[0140] 3. According to the m-projection result, the natural parameters of the target manifold are updated based on formulas (23) and (24). According to the updated natural parameters of the target manifold, the auxiliary manifold parameters are updated based on formulas (25) and (26).

[0141] 5. Repeat the above steps 2 and 3 until the number of iterations T is reached and the target values ​​of the target manifold parameters are obtained.

[0142] 6. Based on formula (27) and formula (28), the target value of the target manifold parameter is converted into the mean and variance of the posterior distribution to obtain the channel estimation result.

[0143] The channel estimation method proposed in the present invention can be subjected to performance tests in three scenarios to demonstrate the technical effects that can be achieved by the present invention.

[0144] Test scenario 1: The number of users is K = 128 and the number of subcarriers is N p =552. The base station is equipped with a 32×1 linear array. During the test, it was compared with the Minimum Mean Square Error (MMSE) algorithm:

[0145]

[0146] During testing, the normalized mean square error is defined as:

[0147]

[0148] Among them G k is the spatial frequency domain channel of user k, is the estimated spatial frequency domain channel of user k.

[0149] In a CDL channel (Clustered Delay Line) model, 128 users were randomly distributed between 0 and 180 degrees, with a delay spread of 30 ns and a channel K factor of 9 dB. The CDL channel is the 5G channel model defined by 3GPP in the 3GPP 38.901 standard. Because user mobility affects the rate of change of adjacent time slot channels, simulations were conducted at different speeds (0, 2, 4, 6, 8, and 10 m / s).

[0150] Figure 3 : is a schematic diagram of the effect of the test scenario 1 provided in an optional embodiment of the present invention, such as Figure 3As shown, the number of iterations, T, is set to 15. The figure shows the performance of each time slot, with the horizontal axis representing the time slot and the vertical axis representing the channel estimation performance (NMSE), expressed in decibels. The red line corresponds to a signal-to-noise ratio of 0 dB, and the blue line corresponds to a signal-to-noise ratio of 20 dB. In each figure, the first time slot has no historical channel information, so the original SIG initialization method (initializing the mean with zero and the variance with D) is used for the initialization of the first time slot. The performance is that of the original SIG algorithm initialization scheme. Figure 3 In (a), the user's moving speed is zero, and the channel in each time slot is exactly the same, so Figure 3 The performance of the channel estimation in (a) gradually improves. Because in this environment, the SIG initialization scheme proposed in the present invention can be regarded as performing T = 15 iterations on the original channel estimation result (the estimation result of the previous time slot) in each time slot, so the performance will improve until convergence. When the moving speed is not zero, the channel of each time slot is different, and the channel estimation performance will tend to be stable. Figure 3 It can be seen that when the SNR is 20dB, the performance of the subsequent time slots (the performance of the initialization scheme proposed in the present invention) can be improved by 3 to 5dB compared to the performance of the first time slot (the performance of the original SIG initialization scheme).

[0151] Figure 4 is a schematic diagram of iterative convergence of test scenario 1 provided according to an optional embodiment of the present invention, Figure 4 Figures a1, a2, and a3 show the specific iterative convergence of the SIG algorithm when the mobile speed is 0 m / s, b1, b2, and b3 show the specific iterative convergence of the SIG algorithm when the mobile speed is 2 m / s, and c1, c2, and c3 show the specific iterative convergence of the SIG algorithm when the mobile speed is 10 m / s. The red line indicates a signal-to-noise ratio of 0 dB; the blue line indicates a signal-to-noise ratio of 20 dB. The dotted line represents the MMSE algorithm; the solid line represents the algorithm provided by the present invention. The adjacent time slot channel change rate is defined as:

[0152]

[0153] Where Q is the number of time slots in the simulation, G q is the channel of the qth time slot.

[0154] Figure 4In groups a and b, due to the low moving speed, the channel correlation between adjacent time slots is strong. When the speed is zero, the channel in each time slot is the same; when the speed is 2m / s, the channel change rate of adjacent time slots is -19.8dB. Therefore, after initialization using the results of the previous time slot, the NMSE has reached below -20dB after one iteration. The performance after 15 iterations is about 5dB higher than the performance of the original initialization scheme (the first time slot). Figure 4 As shown in Group c, when the mobile speed is 10m / s, the channel change rate of adjacent time slots is -6dB. After initialization using the estimation result of the previous time slot, the NMSE can reach below -10dB after one iteration. Finally, after 15 iterations, the performance is about 3dB better than the original initialization scheme.

[0155] Test Scenario 2: This test examines the performance of the initialization scheme when a new user joins the system. The test conditions, including the number of antennas and subcarriers, are the same as those for Test Scenario 1. The system originally had 127 users, and a new user joined the system in the 11th time slot.

[0156] Figure 5 This is a schematic diagram of the effect of the second test scenario provided according to an optional embodiment of the present invention. Figure 5 The channel estimation performance at different speeds is demonstrated when a new user joins the system, where the new user joins in the 11th time slot. Figure 5 Figures a.1, b.1, and c.1 show the average NMSE of users in the system when the SNR is 0 dB (red line) and 20 dB (blue line). As can be seen from the figures, when the initialization scheme provided by the present invention is adopted, the addition of new users has little impact on the channel estimation performance of the entire system, and the performance of the entire system does not deteriorate significantly due to the addition of new users. Figure 5 Figures a.2, b.2, and c.2 show the average NMSE of the original 127 users (red line) and the NMSE of the newly added 128th user (blue line). As can be seen from the figures, the addition of the new user does not significantly interfere with the channel estimates of the existing users. For the new user, since it is essentially initialized with zero mean in the 11th time slot, its performance is slightly lower than that of the remaining 127 users, but still higher than that of a completely zero-mean initialized user (the first time slot in each figure). Starting from the 12th time slot, the 128th user also uses historical channel information for initialization, and its performance quickly converges to that of the remaining 127 users.

[0157] Test Scenario 3: This test simulated performance under different iteration counts, T. Aside from the number of iterations, the rest of the test environment was identical to Test Scenario 1. In the test, the iteration count, T, was set to 3, 4, 5, 8, 12, and 15 times, with a moving speed of 2 m / s. Figure 6This is a schematic diagram of the effect of test scenario three provided according to an optional embodiment of the present invention. Figure 6 The blue line in the middle represents the scheme using zero value initialization, and the red line represents the scheme provided by the present invention for initialization. As can be seen from the figure, the performance of the scheme using historical channel information initialization is higher than that of zero value initialization, and the fewer the number of iterations, the greater the performance gap. For example, when SNR = 20dB, when the number of iterations is 3, the performance of the scheme using historical channel information initialization is 6.8dB higher than the zero value initialization scheme, and when the number of iterations is 15, it is 4.8dB higher. Taking NMSE reaching -20dB as the standard, when SNR = 0dB, 9 iterations are required when using zero value initialization, and 6 iterations are required when using historical channel information initialization; when SNR = 20dB, 9 iterations are required when using zero value initialization, and 4 iterations are required when using historical channel information initialization.

[0158] In summary, the initialization scheme proposed in the present invention can improve the performance of the SIG algorithm with a smaller number of iterations. In other words, the number of iterations can be reduced while maintaining the same channel estimation performance, thereby reducing the complexity of the algorithm.

[0159] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should be aware that the present invention is not limited by the order of the actions described, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0160] Through the description of the above embodiments, those skilled in the art can clearly understand that the channel estimation method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0161] According to an embodiment of the present invention, a channel estimation device for implementing the above-mentioned channel estimation method is also provided. Figure 7 is a structural block diagram of a channel estimation device according to an embodiment of the present invention. Figure 7As shown, the channel estimation device includes: an acquisition module 71, a first determination module 72, a second determination module 77, an update module 74, a third determination module 75 and an estimation module 76. The channel estimation device is described below.

[0162] The acquisition module 71 is configured to acquire an estimated mean of the posterior distribution of the historical channel in a historical time slot as a historical mean.

[0163] The first determination module 72 is connected to the acquisition module 71, and is used to determine the initial mean of the target manifold in the target time slot based on the historical mean, and to determine the initial variance of the target manifold in the target time slot based on the initial mean, wherein the target manifold is used to approximate N nonlinear terms corresponding to the channel estimation, and the target time slot and the historical time slot are adjacent time slots.

[0164] The second determination module 73 is connected to the determination module 72 and is used to determine the initial values ​​of the natural parameters of the target manifold based on the initial mean and initial variance, wherein the natural parameters of the target manifold represent the mean and covariance matrix of the distribution in the target manifold.

[0165] The updating module 74 is connected to the determining module 77 and is used to iteratively update the initial values ​​of the natural parameters of the target manifold to obtain target values ​​of the natural parameters of the target manifold.

[0166] The third determination module 75 is connected to the update module 74 and is used to determine a target mean according to the target value of the natural parameter of the target manifold, and use the target mean as an approximate value of the mean of the posterior distribution of the target channel.

[0167] The estimation module 76 is connected to the third determination module 75 and is configured to perform channel estimation on the target channel in the target time slot according to the target mean value.

[0168] It should be noted that the acquisition module 71, first determination module 72, second determination module 73, update module 74, third determination module 75, and estimation module 76 correspond to steps S201 to S206 in the embodiment. The examples and application scenarios implemented by these modules and the corresponding steps are the same, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above modules, as part of the device, can be run in the computer terminal 10 provided in the embodiment.

[0169] An embodiment of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0170] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the channel estimation method and device in the embodiments of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby implementing the above-mentioned channel estimation method. The memory may include a high-speed random access memory and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely located relative to the processor, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0171] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: obtaining the estimated mean of the posterior distribution of the historical channel in the historical time slot as the historical mean; determining the initial mean of the target manifold in the target time slot based on the historical mean, and determining the initial variance of the target manifold in the target time slot based on the initial mean, wherein the target manifold is used to approximate N nonlinear terms corresponding to the channel estimation, and the target time slot and the historical time slot are adjacent time slots; determining the initial values ​​of the natural parameters of the target manifold based on the initial mean and the initial variance, wherein the natural parameters of the target manifold characterize the mean and covariance matrix of the distribution in the target manifold; iteratively updating the initial values ​​of the natural parameters of the target manifold to obtain target values ​​of the natural parameters of the target manifold; determining the target mean based on the target values ​​of the natural parameters of the target manifold, and using the target mean as an approximate value of the mean of the posterior distribution of the target channel; and performing channel estimation on the target channel in the target time slot based on the target mean.

[0172] Optionally, based on the historical mean, the initial mean of the target manifold in the target time slot is determined, including: obtaining historical object data communicated using the historical channel in the historical time slot, and target object data communicated using the target channel in the target time slot; processing the historical mean based on the historical object data and the target object data to obtain the initial mean.

[0173] Optionally, the historical mean is processed according to the historical object data and the target object data to obtain the initial mean, including: comparing the historical object data and the target object data to obtain a comparison result; obtaining the initial mean based on the comparison result includes at least one of the following: when the historical object data and the target object data are the same, using the historical mean as the initial mean; when there is newly added object data in the target object data, adding a row corresponding to the newly added object data to the historical mean to obtain the initial mean, wherein the newly added object data is data that does not exist in the historical object data, and the historical mean and the initial mean are represented by vectors; when there is leaving object data in the historical object data, deleting a row corresponding to the leaving object data from the historical mean to obtain the initial mean, wherein the leaving object data does not exist in the target object data, and the historical mean and the initial mean are represented by vectors.

[0174] Optionally, determining the initial variance of the target manifold in the target time slot according to the initial mean includes: obtaining target object data communicated using the target channel in the target time slot; and determining the initial variance of the target manifold according to the target object data and the initial mean.

[0175] Optionally, the initial variance of the target manifold in the target time slot is determined based on the target object data and the initial mean, including: determining the energy matrix of the target channel based on the target object data; determining the estimation error generated by adopting the initial mean; and determining the initial variance based on the energy matrix and the estimation error.

[0176] Optionally, the initial values ​​of the natural parameters of the target manifold are iteratively updated to obtain target values ​​of the natural parameters of the target manifold, including: constructing an auxiliary manifold, wherein the auxiliary manifold is used to approximate N-1 nonlinear terms corresponding to the channel estimation; determining the initial values ​​of the natural parameters of the auxiliary manifold based on the initial values ​​of the natural parameters of the target manifold, wherein the natural parameters of the auxiliary manifold characterize the mean and covariance matrix of the distribution in the auxiliary manifold; projecting the auxiliary manifold onto the target manifold to obtain a projection result; updating the natural parameters of the auxiliary manifold based on the projection result, and updating the natural parameters of the target manifold to obtain intermediate values ​​of the natural parameters of the target manifold; using the above method for obtaining the intermediate values ​​of the natural parameters of the target manifold, iteratively updating the intermediate values ​​of the natural parameters of the target manifold to obtain the target values ​​of the natural parameters of the target manifold.

[0177] Optionally, channel estimation is performed on the target channel in the target time slot according to the target mean, including: determining a channel matrix of the target channel according to the target mean; and performing channel estimation on the target channel according to the channel matrix of the target channel.

[0178] According to an embodiment of the present invention, a channel estimation scheme is provided. The estimated mean of the posterior distribution of the historical channel of the adjacent previous time slot is used to determine the initial mean of the target manifold in the target time slot, by obtaining the estimated mean of the posterior distribution of the historical channel in the historical time slot as the historical mean; according to the historical mean, the initial mean of the target manifold in the target time slot is determined, and according to the initial mean, the initial variance of the target manifold in the target time slot is determined, wherein the target manifold is used to approximate N nonlinear terms corresponding to the channel estimation, and the target time slot and the historical time slot are adjacent time slots; based on the initial mean and initial variance, the initial values ​​of the natural parameters of the target manifold are determined, wherein the natural parameters of the target manifold represent the mean and covariance matrix of the distribution in the target manifold. ; Iteratively update the initial values ​​of the natural parameters of the target manifold to obtain the target values ​​of the natural parameters of the target manifold; determine the target mean based on the target values ​​of the natural parameters of the target manifold, and use the target mean as the approximate value of the mean of the posterior distribution of the target channel; perform channel estimation on the target channel within the target time slot based on the target mean, thereby achieving the purpose of reducing the number of iterations from the initial values ​​of the natural parameters of the target manifold to the target values, thereby achieving the technical effect of reducing the computational complexity of channel estimation, and further solving the technical problem of high computational complexity of channel estimation caused by the large number of iterations required for the simplified information geometry channel estimation algorithm in the related technology to achieve better channel estimation performance.

[0179] A person skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a non-volatile storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0180] The embodiment of the present invention further provides a non-volatile storage medium. Optionally, in this embodiment, the non-volatile storage medium can be used to store the program code executed by the channel estimation method provided in the embodiment.

[0181] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0182] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: obtaining an estimated mean of the posterior distribution of a historical channel in a historical time slot as a historical mean; determining an initial mean of the target manifold in the target time slot based on the historical mean, and determining an initial variance of the target manifold in the target time slot based on the initial mean, wherein the target manifold is used to approximate N nonlinear terms corresponding to the channel estimation, and the target time slot and the historical time slot are adjacent time slots; determining initial values ​​of natural parameters of the target manifold based on the initial mean and initial variance, wherein the natural parameters of the target manifold characterize the mean and covariance matrix of the distribution in the target manifold; iteratively updating the initial values ​​of the natural parameters of the target manifold to obtain target values ​​of the natural parameters of the target manifold; determining a target mean based on the target values ​​of the natural parameters of the target manifold, and using the target mean as an approximation of the mean of the posterior distribution of the target channel; and performing channel estimation on the target channel in the target time slot based on the target mean.

[0183] Optionally, based on the historical mean, the initial mean of the target manifold in the target time slot is determined, including: obtaining historical object data communicated using the historical channel in the historical time slot, and target object data communicated using the target channel in the target time slot; processing the historical mean based on the historical object data and the target object data to obtain the initial mean.

[0184] Optionally, the historical mean is processed according to the historical object data and the target object data to obtain the initial mean, including: comparing the historical object data and the target object data to obtain a comparison result; obtaining the initial mean based on the comparison result includes at least one of the following: when the historical object data and the target object data are the same, using the historical mean as the initial mean; when there is newly added object data in the target object data, adding a row corresponding to the newly added object data to the historical mean to obtain the initial mean, wherein the newly added object data is data that does not exist in the historical object data, and the historical mean and the initial mean are represented by vectors; when there is leaving object data in the historical object data, deleting a row corresponding to the leaving object data from the historical mean to obtain the initial mean, wherein the leaving object data does not exist in the target object data, and the historical mean and the initial mean are represented by vectors.

[0185] Optionally, determining the initial variance of the target manifold in the target time slot according to the initial mean includes: obtaining target object data communicated using the target channel in the target time slot; and determining the initial variance of the target manifold according to the target object data and the initial mean.

[0186] Optionally, the initial variance of the target manifold in the target time slot is determined based on the target object data and the initial mean, including: determining the energy matrix of the target channel based on the target object data; determining the estimation error generated by adopting the initial mean; and determining the initial variance based on the energy matrix and the estimation error.

[0187] Optionally, the initial values ​​of the natural parameters of the target manifold are iteratively updated to obtain target values ​​of the natural parameters of the target manifold, including: constructing an auxiliary manifold, wherein the auxiliary manifold is used to approximate N-1 nonlinear terms corresponding to the channel estimation; determining the initial values ​​of the natural parameters of the auxiliary manifold based on the initial values ​​of the natural parameters of the target manifold, wherein the natural parameters of the auxiliary manifold characterize the mean and covariance matrix of the distribution in the auxiliary manifold; projecting the auxiliary manifold onto the target manifold to obtain a projection result; updating the natural parameters of the auxiliary manifold based on the projection result, and updating the natural parameters of the target manifold to obtain intermediate values ​​of the natural parameters of the target manifold; using the above method for obtaining the intermediate values ​​of the natural parameters of the target manifold, iteratively updating the intermediate values ​​of the natural parameters of the target manifold to obtain the target values ​​of the natural parameters of the target manifold.

[0188] Optionally, channel estimation is performed on the target channel in the target time slot according to the target mean, including: determining a channel matrix of the target channel according to the target mean; and performing channel estimation on the target channel according to the channel matrix of the target channel.

[0189] An embodiment of the present invention further provides a computer program product, including a computer program, which implements the steps of the channel estimation method in various embodiments of the present application when the computer program is executed by a processor.

[0190] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0191] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0192] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of the units can be a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0193] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0194] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0195] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, and other media that can store program code.

[0196] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A channel estimation method, characterized in that: include: Obtain the estimated mean of the posterior distribution of the historical channel in the historical time slot as the historical mean; Determining an initial mean of a target manifold in a target time slot based on the historical mean, and determining an initial variance of the target manifold in the target time slot based on the initial mean, wherein the target manifold is used to approximate N nonlinear terms corresponding to channel estimation, and the target time slot and the historical time slot are adjacent time slots; Determining initial values ​​of natural parameters of the target manifold based on the initial mean and the initial variance, wherein the natural parameters of the target manifold represent the mean and covariance matrix of the distribution in the target manifold; Iteratively updating the initial values ​​of the natural parameters of the target manifold to obtain target values ​​of the natural parameters of the target manifold; Determining a target mean according to target values ​​of the natural parameters of the target manifold, and using the target mean as an approximation of the mean of the posterior distribution of the target channel; Channel estimation is performed on the target channel in the target time slot according to the target mean.

2. The method according to claim 1, characterized in that Determining the initial mean value of the target manifold in the target time slot according to the historical mean value includes: Acquire historical object data communicated using the historical channel in the historical time slot, and target object data communicated using the target channel in the target time slot; The historical mean is processed according to the historical object data and the target object data to obtain the initial mean.

3. The method according to claim 2, characterized in that The processing of the historical mean value according to the historical object data and the target object data to obtain the initial mean value includes: comparing the historical object data with the target object data to obtain a comparison result; Obtaining the initial mean based on the comparison result includes at least one of the following: when the historical object data and the target object data are the same, using the historical mean as the initial mean; when there is newly added object data in the target object data, adding a row corresponding to the newly added object data to the historical mean to obtain the initial mean, wherein the newly added object data is data that does not exist in the historical object data, and the historical mean and the initial mean are represented by vectors; when there is departing object data in the historical object data, deleting a row corresponding to the departing object data from the historical mean to obtain the initial mean, wherein the departing object data does not exist in the target object data, and the historical mean and the initial mean are represented by vectors.

4. The method according to claim 1, wherein Determining the initial variance of the target manifold in the target time slot according to the initial mean includes: acquiring target object data communicated using the target channel within the target time slot; An initial variance of the target manifold is determined according to the target object data and the initial mean.

5. The method according to claim 4, characterized in that Determining the initial variance of the target manifold according to the target object data and the initial mean includes: determining an energy matrix of the target channel according to the target object data; determining an estimation error resulting from using the initial mean; The initial variance is determined according to the energy matrix and the estimation error.

6. The method according to claim 1, characterized in that The iterative updating of the initial values ​​of the natural parameters of the target manifold to obtain target values ​​of the natural parameters of the target manifold includes: Constructing an auxiliary manifold, wherein the auxiliary manifold is used to approximate N-1 nonlinear terms corresponding to the channel estimate; Determining initial values ​​of natural parameters of the auxiliary manifold based on initial values ​​of natural parameters of the target manifold, wherein the natural parameters of the auxiliary manifold represent the mean and covariance matrix of the distribution in the auxiliary manifold; Projecting the auxiliary manifold onto the target manifold to obtain a projection result; updating the natural parameters of the auxiliary manifold and the natural parameters of the target manifold according to the projection result, to obtain intermediate values ​​of the natural parameters of the target manifold; The above method for obtaining the intermediate values ​​of the natural parameters of the target manifold is adopted to iteratively update the intermediate values ​​of the natural parameters of the target manifold to obtain the target values ​​of the natural parameters of the target manifold.

7. The method according to any one of claims 1 to 6, characterized in that The performing channel estimation on the target channel in the target time slot according to the target mean value includes: determining a channel matrix of the target channel according to the target mean; Channel estimation is performed on the target channel according to the channel matrix of the target channel.

8. A non-volatile storage medium, characterized in that: The non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the channel estimation method according to any one of claims 1 to 7.

9. A computer device, characterized in that: include: memory and processor, The memory stores a computer program; The processor is configured to execute a computer program stored in the memory, wherein the computer program enables the processor to execute the channel estimation method according to any one of claims 1 to 7 when the program is executed.

10. A computer program product comprising computer instructions, characterized in that The computer instructions are executed by a processor to execute the channel estimation method according to any one of claims 1 to 7.

Images