Wireless channel modeling method, computer device and storage medium

Abstract

The application relates to a wireless channel modeling method, computer equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring a small-scale channel matrix sample; generating a first distribution matrix; the first distribution matrix is an independent and identically distributed complex normal random matrix; generating a first constant matrix, a second constant matrix and a second distribution matrix according to the small-scale channel matrix sample; the first constant matrix is used for reflecting the energy coupling ability of a line-of-sight component, the second constant matrix is used for reflecting the energy coupling ability of a scattering component, and each element of the second distribution matrix is a power of a mutually independent generalized gamma distribution random variable; and acquiring a target channel matrix according to the first constant matrix, the second constant matrix, the first distribution matrix, the second distribution matrix and the small-scale channel matrix sample. The method can realize equivalence of GBSM and CBSM statistical characteristics which are universally applicable to any application scenario.

Description

Technical Field

[0001] This application relates to the field of wireless communication technology, and in particular to a wireless channel modeling method, computer equipment, storage medium, and computer program product. Background Technology

[0002] MIMO (Multiple-Input Multiple-Output) technology refers to the ability to significantly increase the capacity and spectral efficiency of a communication system without increasing bandwidth. It can be defined as the existence of multiple independent channels between the transmitter and receiver, meaning there is sufficient spacing between antenna elements. Therefore, MIMO technology eliminates the correlation between antenna signals, improves signal link performance, and increases data throughput, making it widely used in various wireless communication systems.

[0003] Wireless channel modeling is a core foundation for system design, theoretical analysis, performance evaluation, optimization, and deployment. In MIMO technology research, various channel models are also needed to describe channel characteristics mathematically for analysis and evaluation. Among them, GBSM (Geometrically-based Stochastic Model) is currently recognized as a channel modeling method that can effectively reproduce real-world channel environments. GBSM generally refers to a channel model that generates the channel impulse response based on the physical propagation mechanism of signals, given the positions of the transmitter and receiver and the geometry of the scatterer. This model matches actual channel measurements well, but its complexity is generally high, and the lack of closed-form expressions for channel statistical characteristics hinders mathematical analysis. CBSM (Correlation Based Stochastic Model), on the other hand, does not need to reflect the physical propagation characteristics of signals. Its simpler model and relatively lower complexity facilitate analytical derivation, making it widely used in channel capacity calculation and signal processing algorithm design. However, currently popular CBSM models have stringent assumptions regarding the distribution of independent sub-channel samples, thus limiting their application scenarios.

[0004] Given the important role of wireless channel models in wireless communication research and development, finding a CBSM method equivalent to the statistical characteristics of GBSM channels is an urgent problem to be solved. Summary of the Invention

[0005] Therefore, it is necessary to provide a CBSM wireless channel modeling method, computer equipment, computer-readable storage medium, and computer program product that can achieve equivalent CBSM channel statistical characteristics to GBSM channels, addressing the aforementioned technical problems.

[0006] Firstly, this application provides a wireless channel modeling method. The method includes:

[0007] Obtain small-scale channel matrix samples;

[0008] Generate the first distribution matrix; the first distribution matrix is ​​an independent and identically distributed complex normal random matrix;

[0009] Based on the small-scale channel matrix sample, a first fixed-value matrix, a second fixed-value matrix, and a second distribution matrix are generated. The first fixed-value matrix is ​​used to reflect the energy coupling capability of the line-of-sight component, the second fixed-value matrix is ​​used to reflect the energy coupling capability of the scattering component, and each element of the second distribution matrix is ​​a power of a mutually independent generalized gamma-distributed random variable.

[0010] The target channel matrix is ​​obtained based on the first fixed-value matrix, the second fixed-value matrix, the first distribution matrix, the second distribution matrix, and the small-scale channel matrix sample; the target channel matrix is ​​used to characterize the constructed wireless channel model.

[0011] In one embodiment, generating a first constant matrix, a second constant matrix, and a second distribution matrix based on small-scale channel matrix samples includes:

[0012] Based on the small-scale channel matrix samples, obtain the eigenbase matrices of the transmitter correlation matrix and the receiver correlation matrix;

[0013] The third matrix is ​​obtained based on the eigenbase matrix of the transmitter correlation matrix, the eigenbase matrix of the receiver correlation matrix, and the small-scale channel matrix sample.

[0014] Based on the expectation of the third matrix, obtain the first constant matrix;

[0015] The second constant value matrix is ​​obtained based on the standard deviation of the third matrix.

[0016] In one embodiment, generating a first constant matrix, a second constant matrix, and a second distribution matrix based on small-scale channel matrix samples includes:

[0017] The distribution of small-scale channel matrix samples is fitted to obtain the probability distribution model of small-scale channel matrix samples;

[0018] Based on the probability distribution model, determine the base and exponent of the power corresponding to each element in the second distribution matrix; the base of the power follows a generalized gamma distribution.

[0019] In one embodiment, the probability distribution model includes a complex Gaussian distribution, a complex Student's distribution, and a complex Laplace distribution;

[0020] Based on the probability distribution model, the base and exponent of the power corresponding to each element in the second distribution matrix are determined as follows:

[0021] When the probability distribution model is a complex Gaussian distribution, the exponent of the power is 0;

[0022] When the probability distribution model is a complex student distribution, the exponent of the power is the first value, and the base of the power is obtained by fitting the distribution of the small-scale channel matrix samples.

[0023] When the probability distribution model is a complex Laplace distribution, the exponent of the power is the second value, and the base of the power is determined according to the first and second parameters; the first parameter is obtained from the peak coefficient estimated by the discrete samples of the third matrix, and the second parameter is a preset shape parameter.

[0024] In one embodiment, after generating a first constant matrix, a second constant matrix, and a second distribution matrix based on small-scale channel matrix samples, the process includes:

[0025] Based on the first distribution matrix and the second distribution matrix, obtain a mixed-distribution random sample;

[0026] Based on small-scale channel matrix samples, obtain the target probability distribution function;

[0027] Based on the probability distribution function of the mixed-distribution random samples, obtain the fitted probability distribution function;

[0028] Obtain the degree of difference between the target probability distribution function and the fitted probability distribution function;

[0029] Based on a mixed-distribution random sample, the first and second parameters are adjusted according to the degree of difference to optimize the mixed-distribution random sample.

[0030] In one embodiment, optimizing the mixed-distribution random samples by adjusting a first parameter and a second parameter according to the degree of dissimilarity includes:

[0031] When the difference is greater than the threshold, the first parameter and the second parameter are adjusted within the preset range until the difference of the mixed distribution random samples corresponding to the adjusted first parameter and the second parameter meets the condition of not being greater than the threshold.

[0032] For each degree of difference that meets the conditions, a mixed distribution random sample is obtained, which is determined by the first and second parameters corresponding to the degree of difference.

[0033] In one embodiment, obtaining the difference between the target probability distribution function and the fitted probability distribution function includes:

[0034] Based on the target probability distribution function and the fitted probability distribution function, obtain the forward KL divergence and the inverse KL divergence;

[0035] The difference is obtained by summing the forward KL divergence and the backward KL divergence.

[0036] Secondly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method.

[0037] Thirdly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0038] Thirdly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method.

[0039] The aforementioned wireless channel modeling method, computer equipment, storage medium, and computer program products, firstly, compared to GBSM, do not require analysis of the physical propagation characteristics of the signal, resulting in a simpler model and the ability to form a deterministic expression for the target channel matrix, which is beneficial for mathematical analysis. Secondly, based on the probability distribution model of small-scale channel matrix samples, this application proposes a hybrid distribution random sample synthesized from a generalized gamma power distribution and a complex normal distribution. The elements of the hybrid distribution random sample are independent and can follow the same or different types of distributions. Furthermore, this application introduces line-of-sight components and scattering components, making it more universally applicable to fitting channel samples of arbitrary distribution forms compared to CBSM. It also achieves a good approximation of GBSM in statistical characteristics, especially in large-scale MIMO systems, significantly improving statistical accuracy. Therefore, this application proposes a universal, high-precision channel modeling method that considers the distribution characteristics of actual wireless channel samples, achieving statistical equivalence between GBSM and CBSM applicable to any scenario. Attached Figure Description

[0040] Figure 1 This is a diagram illustrating the application environment of a wireless channel modeling method in one embodiment.

[0041] Figure 2 This is a flowchart illustrating a wireless channel modeling method in one embodiment;

[0042] Figure 3 This is a schematic diagram of the process for generating mixed-distribution random samples in one embodiment;

[0043] Figure 4 This is a schematic diagram comparing the probability density distribution of channel capacity in a propagation scenario where the Rice K-factor is 9dB in one embodiment.

[0044] Figure 5 This is a schematic diagram comparing the probability density distribution of feature values ​​in a propagation scenario where the Rice K-factor is 9dB in one embodiment.

[0045] Figure 6 This is a schematic diagram comparing the probability density distribution of channel capacity in a propagation scenario where the Rice K-factor is -100dB in one embodiment.

[0046] Figure 7 This is a schematic diagram comparing the probability density distribution of feature values ​​in a propagation scenario where the Rice K-factor is -100dB in one embodiment.

[0047] Figure 8 This is a structural block diagram of a wireless channel modeling device in one embodiment;

[0048] Figure 9 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0050] The wireless channel modeling method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be located in the cloud or on other network servers. The data storage system can store channel data, etc. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, etc. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0051] In the field of MIMO technology, GBSM and CBSM methods are commonly used for channel modeling. GBSM can match actual channel measurements well, but its complexity is generally high, its channel modeling efficiency is lower than CBSM, and its channel statistical characteristics lack closed-form expressions, which is not conducive to mathematical analysis. CBSM, on the other hand, has a relatively simple model, but its application scenarios have many limitations, making it unable to reproduce more complex channels. Furthermore, the accuracy of the model decreases as the scale of the communication system increases, often resulting in an overestimation of channel capacity. Common CBSM channel models include the Weichselberger Channel Model and the Jointly Correlated Channel Model (JCCM).

[0052] The Vieux Model models joint correlation characteristics by describing the average coupling between the eigenmodes of the two link ends. A necessary and sufficient condition for this model to hold is that the receiver's eigenbase is independent of the transmit weights, and vice versa. The model's input parameters can be extracted from a random channel sample set through simulation or measurement. The model's basic assumptions can be explained by the principles of radio wave propagation and it is backward compatible with the Kronecker channel model and the virtual channel representation model. However, the Vieux Model is based on complex Gaussian random matrices, therefore it can only achieve accurate statistical channel modeling under the condition of independent Rayleigh fading channels.

[0053] JCCM is an improved CBSM model based on the Wechsler model, which introduces a constant matrix to represent the line-of-sight (LOS) component. The model's input parameters can be extracted from a random channel sample set through simulation or measurement. Its basic assumptions can be explained by radio wave propagation principles, and it is backward compatible with related random channel models such as the Kronecker channel model, virtual channel representation model, and Wechsler model. While JCCM relaxes the requirements for the distribution of random matrix elements to some extent, not mandating that the random matrix elements (each independent sub-channel) follow a complex Gaussian distribution, it still requires that they be independent and identically distributed.

[0054] In summary, the main problems with CBSM are as follows:

[0055] 1. Currently, there may be discrepancies between the statistical characteristics of major CBSM and GBSM or actual channels, and this gap will widen as the scale of MIMO increases. Given the application of large-scale MIMO arrays in next-generation wireless communication technologies, this gap cannot be ignored and may lead to serious deviations in system performance evaluation.

[0056] 2. Currently, most CBSMs have relatively strict assumptions about the distribution of independent sub-channel samples, which limits their application scenarios.

[0057] 3. Currently, the main CBSMs are only applicable to cluster-level and link-level channel modeling, but do not support system-level channel modeling.

[0058] Therefore, given the important role of wireless channel models in wireless communication research and development, finding a CBSM method equivalent to the statistical characteristics of GBSM channels is an urgent problem to be solved.

[0059] To address the above problems, in one embodiment, such as Figure 2 As shown, a wireless channel modeling method is provided, with a universal CBSM implementation called GJCCM (Generalized Jointly Correlated Channel Model). This method is applied to... Figure 1 Taking terminal 102 as an example, the explanation includes the following steps:

[0060] Step 202: Obtain small-scale channel matrix samples.

[0061] Both the transmitting and receiving antennas are omnidirectional antennas with arbitrary polarization. Random channel samples are collected using Monte Carlo simulation or actual channel measurements to obtain a sample of size N. r ×N t Channel matrix sample H × N raw N r N represents the number of user antennas at the receiving end. t N represents the number of transmitting antennas, and N represents the number of Monte Carlo simulations or the number of actual channel measurement snapshots (samples).

[0062] In the nth (n=1,…,N) Monte Carlo simulation or channel measurement, the channel gain of the receiving user is denoted as... The channel matrix is Then the receiving user corresponds to the small-scale channel matrix sample in the nth simulation or measurement. The formula for calculation is:

[0063]

[0064] After N Monte Carlo simulations or channel measurements, the normalized channel matrix obtained by the receiving user is of size N. r ×N t Small-scale channel matrix H × N samples SS .

[0065] Step 204: Generate the first distribution matrix; the first distribution matrix is ​​an independent and identically distributed complex normal random matrix.

[0066] Step 206: Based on the small-scale channel matrix sample, generate a first fixed-value matrix, a second fixed-value matrix, and a second distribution matrix; the first fixed-value matrix is ​​used to reflect the energy coupling capability of the line-of-sight component, the second fixed-value matrix is ​​used to reflect the energy coupling capability of the scattering component, and each element of the second distribution matrix is ​​a power of a mutually independent generalized gamma-distributed random variable.

[0067] Based on the small-scale channel matrix sample, obtain the first fixed matrix that reflects the energy coupling capability of the line-of-sight component and the second fixed matrix that reflects the energy coupling capability of the scattering component.

[0068] By incorporating line-of-sight components and scattering components into the channel modeling framework, and adjusting the data of the first and second fixed-value matrices according to the actual channel, the requirements for the distribution of random matrix elements are relaxed to some extent, thus expanding the application scenarios of this embodiment.

[0069] In one embodiment, a probability distribution model of a small-scale channel matrix sample is obtained, and a second distribution matrix is ​​generated based on the probability distribution model.

[0070] The elements of the second distribution matrix appear in the form of exponents, and the base of the exponent follows a generalized gamma distribution. The generalized gamma distribution contains several parameters, and the parameters and the exponent of the exponent are determined by the probability distribution model.

[0071] Step 208: Obtain the target channel matrix based on the first fixed value matrix, the second fixed value matrix, the first distribution matrix, the second distribution matrix, and the small-scale channel matrix sample; the target channel matrix is ​​used to characterize the constructed wireless channel model.

[0072] The overall matrix of the channel scattering environment is represented by the first constant matrix, the second constant matrix, the first distribution matrix, and the second distribution matrix. Based on small-scale channel matrix samples, the spatial characteristic basis matrices of the receiver and the transmitter are obtained. Combined with the overall matrix of the channel scattering environment, the final target channel matrix is ​​obtained, thus completing the construction of the channel model.

[0073] The spatial characteristic basis matrix of the receiver and the spatial characteristic basis matrix of the transmitter are determined by the antenna configuration of the receiver and transmitter, and are independent of the scattering environment.

[0074] The aforementioned wireless channel modeling methods yield concise and clear channel model expressions, and are universally applicable to fitting channel samples in any scenario, including arbitrary sample distributions and antenna configurations. Furthermore, based on mixed-distribution random samples jointly defined by the generalized gamma power distribution and the complex normal distribution, they achieve good approximation of the statistical characteristics of GJCCM and GBSM, demonstrate good performance in joint correlation channel modeling at any MIMO scale, and are also suitable for cluster-level, link-level, and system-level channel modeling.

[0075] In one embodiment, step 204 includes: generating a complex normal distribution matrix in which the real and imaginary parts of each element are independently and identically distributed, with a mean of 0 and a variance of 1 / 2, which is the first distribution matrix G. iid The first distribution matrix G iid It can be a fixed value.

[0076] In one embodiment, step 206 includes: obtaining the eigenbase matrices of the transmitter correlation matrix and the receiver correlation matrix based on the small-scale channel matrix samples; obtaining a third matrix based on the eigenbase matrices of the transmitter correlation matrix, the receiver correlation matrix, and the small-scale channel matrix samples; obtaining a first fixed-value matrix based on the expectation of the third matrix; and obtaining a second fixed-value matrix based on the standard deviation of the third matrix.

[0077] Among them, the eigenbase matrix of the transmitter correlation matrix is ​​the spatial eigenbase matrix of the transmitter, and the eigenbase matrix of the receiver correlation matrix is ​​the spatial eigenbase matrix of the receiver. The eigenbase matrix U of the transmitter correlation matrix t The eigenbase matrix U of the correlation matrix at the receiving end r The calculation process is as follows:

[0078]

[0079]

[0080] Among them, H SS For small-scale channel matrix samples, R t R is the correlation matrix of the transmitter. r For the correlation matrix at the receiving end, Λ t For corresponding to R t eigenvalue diagonal matrix, Λ r For corresponding to R r The eigenvalue diagonal matrix is ​​given by E(·), which is the expected function.

[0081] The third matrix is ​​obtained based on the eigenbase matrices of the transmitter correlation matrix, the eigenbase matrices of the receiver correlation matrix, and small-scale channel matrix samples. The calculation process is as follows:

[0082]

[0083] Based on the expectation of the third matrix, the first constant matrix D is obtained, and its calculation process is as follows:

[0084]

[0085] Based on the standard deviation of the third matrix, the second constant matrix M is obtained, and its calculation process is as follows:

[0086]

[0087] Where i and j represent the third matrix respectively. The row and column indices are given, and var(·) is the variance function.

[0088] Solve for the third matrix, the first constant matrix, and the second constant matrix to prepare data for obtaining the target channel matrix later.

[0089] In one embodiment, step 206 further includes: performing distribution fitting on the small-scale channel matrix samples to obtain a probability distribution model of the small-scale channel matrix samples; and determining the base and exponent of the power corresponding to each element in the second distribution matrix based on the probability distribution model; wherein the base of the power follows a generalized gamma distribution.

[0090] The second distribution matrix is ​​used in R. ind This indicates that its element expression can be represented as:

[0091]

[0092] Where i and j represent R respectively ind row index and column index; r i,j The base of the power is r, which follows a generalized gamma distribution, i.e., r i,j ~Γ(α i,j ,β i,j ,γ i,j ,μ i,j ), α i,j ,β i,j ,γ i,j ,μ i,j They represent [R] ind ] i,j The corresponding first shape parameter, scale parameter, second shape parameter, and position parameter, and α i,j ,β i,j ,γ i,j All are definite positive real numbers, μ i,j To determine the real number; p i,j The exponent is the power of the number of powers.

[0093] The second distribution matrix is ​​determined based on the probability distribution model obtained by fitting the sample distribution of the small-scale channel matrix, thereby improving the accuracy of channel simulation.

[0094] In one embodiment, the probability distribution model includes a complex Gaussian distribution, a complex Student's distribution, and a complex Laplace distribution.

[0095] In this embodiment, based on shape characteristics, it is first determined whether the probability distribution model conforms to a complex Gaussian distribution or a complex Student's distribution. When the probability distribution model is neither a complex Gaussian distribution nor a complex Student's distribution, it is considered that the probability distribution model is more approximately similar to a complex Laplace distribution, and the channel model is further constructed according to the case of a complex Laplace distribution.

[0096] These three models can cover the vast majority of real-world channels. Therefore, channel modeling can be performed quickly and conveniently using only these three models, effectively reducing the complexity of analysis and thus improving modeling efficiency.

[0097] In one embodiment, determining the base and exponent of the power corresponding to each element in the second distribution matrix, based on the probability distribution model, includes:

[0098] When the probability distribution model is a complex Gaussian distribution, the exponent of the power is 0; when the probability distribution model is a complex Student's distribution, the exponent of the power is the first value, and the base of the power is obtained by fitting the distribution of the small-scale channel matrix samples; when the probability distribution model is a complex Laplace distribution, the exponent of the power is the second value, and the base of the power is determined by the first parameter and the second parameter; the first parameter is obtained by estimating the peak coefficient of the discrete samples of the third matrix, and the second parameter is a preset shape parameter.

[0099] In one embodiment, after generating a first fixed-value matrix, a second fixed-value matrix, and a second distribution matrix based on small-scale channel matrix samples, the method includes: obtaining mixed-distribution random samples based on the first distribution matrix and the second distribution matrix.

[0100] The first and second distribution matrices are of the same order. A mixed-distribution random sample is obtained using the Hadamard product of the first and second distribution matrices. Define the mixed-distribution random sample H. ind The expression is H ind =R ind ⊙G iid , where R ind Let G represent the second distribution matrix. iid Let represent the first distribution matrix. The elements of a mixed-distribution random sample are independent and can follow the same or different types of distributions. Since the second distribution matrix is ​​generated based on a probability distribution model, different probability distribution models correspond to different second distribution matrices; therefore, the mixed-distribution random sample also varies depending on the probability distribution model.

[0101] When the probability distribution model is a complex Gaussian distribution, the exponent of the power is 0, and the base of the power can be any non-zero value, meaning that all elements of the second distribution matrix are 1. In this case, the mixed distribution random sample is equal to the first distribution matrix G. iid .

[0102] In one embodiment, when the probability distribution model is a complex student distribution, according to the second distribution matrix R... ind expression Exponent of power That is, the first value is The base r of the power i,j The second shape parameter γ in i,j The value is 1. At this time, the mixed distribution random sample H... ind The elements in are Where g i,j The first distribution matrix G iid The elements. A mixed-distribution random sample is transformed into a complex student distribution, i.e., [H... ind ] i,j ~t(0,σ i,j,v i,j ), σ i,j and v i,j α represents the parameter of the student distribution, which is obtained by fitting the distribution to a small-scale channel matrix sample. i,j ,β i,j With σ i,j ,v i,j The correspondence is as follows:

[0103]

[0104] According to σ i,j and v i,j α can be obtained i,j ,β i,j γ i,j =1, μ i,j It is meaningless and can be any number, thus yielding the second distribution matrix R. ind According to the distribution of returning students, the corresponding second distribution matrix R ind and the first distribution matrix G iid Obtain a mixed distribution random sample H ind .

[0105] In one embodiment, when the probability distribution model is a complex student distribution, a mixed distribution random sample H can be obtained directly by performing distribution fitting. ind .

[0106] In one embodiment, when the probability distribution model is a complex Laplace distribution, the second distribution matrix R ind expression The exponent p of the power i,j =1, meaning the second value is 1. The base r of the exponent. i,j The parameters in the equations can be obtained from the following system of equations:

[0107]

[0108] Where Gamma(·) represents the gamma function (i.e., Euler's second integral), k i,j The first parameter is s, which takes the value of the peak coefficient of the discrete sample estimation of the third matrix. i,j The second parameter, introduced by α, is used to ensure a unique solution. i,j and γ i,j The shape parameter is defined by the product.

[0109] By obtaining the parameters of the generalized gamma distribution matrix from the system of equations, we can obtain the base of the power. Combined with the exponent of the power, we can then obtain the second distribution matrix R corresponding to the complex Laplace distribution. ind Furthermore, based on the second distribution matrix R corresponding to the complex Laplace distribution... ind and the first distribution matrix Giid Obtain a mixed distribution random sample H ind .

[0110] By calculating the second distribution matrix for different probability distribution models, a target channel matrix with higher simulation accuracy can be obtained.

[0111] In one embodiment, after generating the first fixed-value matrix, the second fixed-value matrix, and the second distribution matrix based on the small-scale channel matrix samples, the method further includes: obtaining a target probability distribution function based on the small-scale channel matrix samples; obtaining a fitted probability distribution function based on the probability distribution function of the mixed-distribution random samples; obtaining the difference between the target probability distribution function and the fitted probability distribution function; and adjusting the first and second parameters based on the difference to optimize the mixed-distribution random samples. Since the first parameter is an estimate based on discrete samples, it inevitably has biases, therefore, the second distribution matrix needs further optimization to obtain more accurate mixed-distribution random samples.

[0112] In one embodiment, optimizing the mixed-distribution random sample by adjusting the first parameter and the second parameter according to the degree of difference includes: when the degree of difference is greater than a threshold, adjusting the first parameter and the second parameter within a preset range until the degree of difference of the mixed-distribution random sample corresponding to the adjusted first parameter and the second parameter meets the condition of not being greater than the threshold; for the degree of difference that meets the condition, obtaining the mixed-distribution random sample determined by the first parameter and the second parameter corresponding to the degree of difference.

[0113] By comparing the fitted probability distribution function with the target probability distribution function, when the difference is within a threshold range, it indicates that the mixed-distribution random sample is close to the small-scale channel matrix sample in terms of probability distribution. Through this comparison, more accurate first and second parameters are obtained. Based on these parameters, a more accurate mixed-distribution random sample can be obtained, ultimately leading to a target channel matrix with a good fit.

[0114] In one embodiment, obtaining the difference between the target probability distribution function and the fitted probability distribution function includes: obtaining the forward KL divergence and the backward KL divergence based on the target probability distribution function and the fitted probability distribution function; and obtaining the difference based on the sum of the forward KL divergence and the backward KL divergence.

[0115] Specifically, this embodiment calculates the degree of difference using a cost function, which is:

[0116]

[0117] Among them, D KLDenotes the KL divergence (i.e., Kullback-Leibler divergence); p(·) and q(·) are the target probability distribution function and the fitted probability distribution function, respectively, and D KL (p|q) is the forward KL divergence, D KL (q|p) represents the inverse KL divergence; L is the number of sample points used for alignment, x l Let $l$ be the value of the $l$-th random variable, and $cost$ is used to measure the degree of variability.

[0118] When the cost value exceeds the threshold, the probability distribution function of the mixed-distribution random sample differs significantly from that of the small-scale channel matrix sample. Therefore, the first and second parameters need to be adjusted within a preset range to obtain a new mixed-distribution random sample. Based on the new mixed-distribution random sample, the difference in probability distribution function between the new mixed-distribution random sample and the small-scale channel matrix sample is recalculated. This process is repeated until the cost value is less than or equal to the threshold. At this point, a mixed-distribution random sample whose difference meets the condition is obtained; this is the optimized mixed-distribution random sample, and the target channel matrix obtained based on it is the final result.

[0119] After constructing the first and second distribution matrices, the channel model expression is as follows:

[0120]

[0121] Among them, H GJCCM For the target channel matrix, U r U is the eigenbase matrix of the correlation matrix at the receiving end. t Let H be the eigenbase matrix of the correlation matrix at the transmitting end, D be the first constant matrix, M be the second constant matrix, and H be the eigenbase matrix. ind For a mixed-distribution random sample, ⊙ represents the Hadamard product. M⊙H ind D+M⊙H represents the gain of each sub-channel in the scattering environment. ind This represents the overall matrix of the channel scattering environment. Where H... ind According to formula H ind =R ind ⊙G iid Obtain, R ind Let G represent the second distribution matrix. iid Let represent the first distribution matrix.

[0122] In one embodiment, such as Figure 3The diagram illustrates the process of obtaining a mixed-distribution random sample. The main steps are generating a first distribution matrix and a second distribution matrix, which can be considered sequentially. Generating the second distribution matrix includes the following steps: obtaining the probability distribution model of the small-scale channel matrix sample and determining whether the probability distribution model is a complex Gaussian distribution. If yes, the exponent of the power is 0, and a second distribution matrix corresponding to the complex Gaussian distribution is generated accordingly; if not, determining whether the probability distribution model is a complex Student's distribution. If yes, the exponent of the power is -1 / 2, and the base of the power is obtained by fitting the distribution of the small-scale channel matrix sample, and a second distribution model corresponding to the complex Student's distribution is generated accordingly; if not, determining it is a complex Laplace distribution, the exponent of the power is 1, and the base of the power is determined according to the first and second parameters, and a second distribution matrix corresponding to the complex Laplace distribution is generated accordingly. The mixed-distribution random sample is synthesized using the Hadamard product of the first and second distribution matrices. Based on the mixed-distribution random sample, the difference between the target probability distribution function and the fitted probability distribution function is obtained. The difference is determined to be greater than a threshold. If yes, the first and second parameters are adjusted, and the second distribution matrix is ​​regenerated. Repeat the process of acquiring and judging mixed-distribution random samples until the difference meets the condition of not exceeding a threshold. Output the corresponding mixed-distribution random samples whose difference meets the condition for subsequent acquisition of the target channel matrix.

[0123] In one embodiment, the wireless channel modeling method includes the following steps:

[0124] Obtain small-scale channel matrix samples.

[0125] Generate the first distribution matrix.

[0126] Based on the small-scale channel matrix samples, obtain the eigenbase matrices of the transmitter correlation matrix and the receiver correlation matrix.

[0127] The third matrix is ​​obtained based on the eigenbase matrix of the transmitter correlation matrix, the eigenbase matrix of the receiver correlation matrix, and the small-scale channel matrix sample.

[0128] Based on the expectation of the third matrix, obtain the first constant matrix.

[0129] The second constant value matrix is ​​obtained based on the standard deviation of the third matrix.

[0130] Distribution fitting is performed on small-scale channel matrix samples to obtain probability distribution models for the small-scale channel matrix samples. The probability distribution models include complex Gaussian distribution, complex Student distribution, and complex Laplace distribution.

[0131] Based on the probability distribution model, the base and exponent of the power corresponding to each element in the second distribution matrix are determined; the base of the power follows a generalized gamma distribution. When the probability distribution model is a complex Gaussian distribution, the exponent of the power is 0. When the probability distribution model is a complex Student's distribution, the exponent of the power is the first value, and the base of the power is obtained by fitting the distribution of the small-scale channel matrix samples. When the probability distribution model is a complex Laplace distribution, the exponent of the power is the second value, and the base of the power is determined by the first and second parameters. The first parameter is obtained from the peak coefficient estimated from the discrete samples of the third matrix, and the second parameter is a preset shape parameter.

[0132] The second distribution matrix is ​​generated based on the base and exponent of the power.

[0133] Based on the first distribution matrix and the second distribution matrix, obtain a mixed distribution random sample.

[0134] The target probability distribution function is obtained based on small-scale channel matrix samples.

[0135] Obtain the fitted probability distribution function based on the probability distribution function of the mixed distribution random sample.

[0136] Based on the target probability distribution function and the fitted probability distribution function, obtain the forward KL divergence and the inverse KL divergence.

[0137] The difference is obtained by summing the forward KL divergence and the backward KL divergence.

[0138] When the difference is greater than the threshold, the first parameter and the second parameter are adjusted within a preset range until the difference of the mixed distribution random samples corresponding to the adjusted first parameter and the second parameter meets the condition of not being greater than the threshold.

[0139] For each degree of difference that meets the conditions, a mixed distribution random sample is obtained, which is determined by the first and second parameters corresponding to the degree of difference.

[0140] The target channel matrix is ​​obtained based on the first fixed-value matrix, the second fixed-value matrix, the mixed-distribution random sample, and the small-scale channel matrix sample.

[0141] The advantages of this embodiment compared to the prior art are:

[0142] 1. A GJCCM is proposed, which forms a concise model expression formula that is easy to operate and convenient for mathematical analysis.

[0143] 2. In terms of the implementation of GJCCM, a novel hybrid distribution based on the generalized gamma power distribution and the complex normal distribution is proposed for the first time, which can be universally applied to the fitting of channel samples with arbitrary distribution forms.

[0144] 3. In terms of GJCCM implementation, based on the mixed distribution random samples defined by the generalized gamma power distribution and the complex normal distribution, it can achieve a good approximation of the statistical characteristics of GJCCM and GBSM.

[0145] 4. In terms of applicability, GJCCM can be universally applied to joint correlation channel modeling of any MIMO scale.

[0146] 5. In terms of applicability, GJCCM can be universally applied to channel modeling in any scenario.

[0147] 6. In terms of applicability, GJCCM is backward compatible with JCCM, the Wechsler model, and other existing CBSM models, for example:

[0148] (1) When the elements in the mixed distribution random sample are independent and identically distributed, the proposed GJCCM can be degenerated into JCCM;

[0149] (2) When each element in the mixed distribution random sample is independently and identically distributed in the standard complex normal distribution, and D = 0 (NLOS channel), the proposed GJCCM degenerates into the Wechsler model;

[0150] (3) Based on b, if the joint correlation between the transmitting and receiving ends is further ignored, the GJCCM degenerates into the Kronecher model.

[0151] The GJCCM proposed in this invention achieves excellent agreement with the statistical characteristics of GBSM channel matrix samples (generated by the classic QuaDRiGa simulation platform), realizing the equivalence of CBSM and GBSM channel statistical characteristics. It is applicable to channel modeling in any scenario and at any MIMO scale, and is backward compatible with existing CBSM methods. The following simulation experiments will verify the performance of GJCCM:

[0152] The simulation parameter settings are shown in Table 1:

[0153] Table 1. Simulation parameter settings for UMi propagation scenario

[0154]

[0155]

[0156] like Figure 4 and Figure 5 The figure shows a simulation of the LOS propagation scenario in UMi (Urban Microcell) under the condition that the Rice K factor is fixed (K = 9dB). Figure 4 and Figure 5Comparison diagrams are presented of the channel capacity and corresponding eigenvalue probability density distributions of GBSM, JCCM, and the GJCCM proposed in this invention under the UMi LOS propagation scenario. From Figure 4 It is evident that, due to the assumption by JCCM that all independent random sub-channels follow a complex Gaussian distribution, JCCM overestimates the channel capacity corresponding to GBSM, and JCCM is no longer applicable at this point. In contrast, the curve of GJCCM achieves a better match with GBSM. Figure 5 This indicates that, compared to JCCM, the eigenvalues ​​of the channel matrix of GJCCM match the GBSM results better, thus achieving a closer approximation of the channel statistical characteristics of CBSM and GBSM.

[0157] like Figure 6 and Figure 7 The figure shows a simulation of the UMi NLOS propagation scenario under the condition that the Rice K factor is a fixed value (K = -100dB). Figure 6 and Figure 7 Comparison diagrams are presented of the channel capacity and corresponding eigenvalue probability density distributions of GBSM, JCCM, and the GJCCM proposed in this invention under the UMi NLOS propagation scenario. From Figure 6 It is evident that, similarly, JCCM overestimates the channel capacity of GBSM, while the curve of GJCCM fits GBSM well. Figure 7 This demonstrates that GJCCM can better match the eigenvalue distribution than JCCM, thus achieving a better approximation of the channel statistical characteristics.

[0158] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0159] Based on the same inventive concept, this application also provides a wireless channel modeling apparatus for implementing the wireless channel modeling method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more wireless channel modeling apparatus embodiments provided below can be found in the limitations of the wireless channel modeling method described above, and will not be repeated here.

[0160] In one embodiment, such as Figure 8 As shown, a wireless channel modeling device is provided, including: a sample acquisition module 802, a first generation module 804, a second generation module 806, and a model generation module 808, wherein:

[0161] The sample acquisition module 802 is used to acquire small-scale channel matrix samples.

[0162] The first generation module 804 is used to generate the first distribution matrix; the first distribution matrix is ​​an independent and identically distributed complex normal random matrix.

[0163] The second generation module 806 is used to generate a first fixed-value matrix, a second fixed-value matrix, and a second distribution matrix based on the small-scale channel matrix sample. The first fixed-value matrix is ​​used to reflect the energy coupling capability of the line-of-sight component, the second fixed-value matrix is ​​used to reflect the energy coupling capability of the scattering component, and each element of the second distribution matrix is ​​a power of a mutually independent generalized gamma-distributed random variable.

[0164] The model generation module 808 is used to obtain the target channel matrix based on the first fixed value matrix, the second fixed value matrix, the first distribution matrix, the second distribution matrix, and the small-scale channel matrix sample; the target channel matrix is ​​used to characterize the constructed wireless channel model.

[0165] The second generation module 806 is further configured to obtain the characteristic basis matrices of the transmitter correlation matrix and the receiver correlation matrix based on the small-scale channel matrix samples; obtain a third matrix based on the characteristic basis matrices of the transmitter correlation matrix, the receiver correlation matrix, and the small-scale channel matrix samples; obtain a first fixed-value matrix based on the expectation of the third matrix; and obtain a second fixed-value matrix based on the standard deviation of the third matrix.

[0166] The second generation module 806 is also used to perform distribution fitting on the small-scale channel matrix samples to obtain the probability distribution model of the small-scale channel matrix samples; based on the probability distribution model, it determines the base and exponent of the power corresponding to each element in the second distribution matrix; the base of the power follows a generalized gamma distribution.

[0167] The second generation module 806 is also used to: when the probability distribution model is a complex Gaussian distribution, the exponent of the power is 0; when the probability distribution model is a complex Student's distribution, the exponent of the power is a first value, and the base of the power is obtained by fitting the distribution of the small-scale channel matrix samples; when the probability distribution model is a complex Laplace distribution, the exponent of the power is a second value, and the base of the power is determined by the first parameter and the second parameter; the first parameter is obtained by estimating the peak coefficient of the discrete samples of the third matrix, and the second parameter is a preset shape parameter.

[0168] The second generation module 806 is further configured to, after generating a first fixed-value matrix, a second fixed-value matrix, and a second distribution matrix based on the small-scale channel matrix samples, obtain mixed-distribution random samples based on the first distribution matrix and the second distribution matrix; obtain a target probability distribution function based on the small-scale channel matrix samples; obtain a fitted probability distribution function based on the probability distribution function of the mixed-distribution random samples; obtain the difference between the target probability distribution function and the fitted probability distribution function; and adjust the first parameter and the second parameter based on the difference based on the mixed-distribution random samples to optimize the mixed-distribution random samples.

[0169] The second generation module 806 is also used to adjust the first parameter and the second parameter within a preset range when the difference is greater than the threshold, until the difference of the mixed distribution random sample corresponding to the adjusted first parameter and the second parameter meets the condition of not being greater than the threshold; for the difference that meets the condition, a mixed distribution random sample determined by the first parameter and the second parameter corresponding to the difference is obtained.

[0170] The second generation module 806 is also used to obtain the forward KL divergence and the backward KL divergence based on the target probability distribution function and the fitted probability distribution function; and to obtain the difference degree based on the sum of the forward KL divergence and the backward KL divergence.

[0171] Each module in the aforementioned wireless channel modeling device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0172] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 9 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores channel data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a wireless channel modeling method.

[0173] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 9 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a wireless channel modeling method.

[0174] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0175] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in all of the above method embodiments.

[0176] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in all of the above method embodiments.

[0177] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the steps in all of the above method embodiments.

[0178] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0179] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0180] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A wireless channel modeling method, characterized in that, The method includes: Obtain small-scale channel matrix samples; Generate a first distribution matrix; the first distribution matrix is ​​an independent and identically distributed complex normal random matrix; Based on the small-scale channel matrix sample, a first fixed-value matrix, a second fixed-value matrix, and a second distribution matrix are generated; the first fixed-value matrix is ​​used to reflect the energy coupling capability of the line-of-sight component, the second fixed-value matrix is ​​used to reflect the energy coupling capability of the scattering component, and each element of the second distribution matrix is ​​a power of a mutually independent generalized gamma-distributed random variable. The target channel matrix is ​​obtained based on the first fixed value matrix, the second fixed value matrix, the first distribution matrix, the second distribution matrix, and the small-scale channel matrix sample; the target channel matrix is ​​used to characterize the constructed wireless channel model.

2. The method according to claim 1, characterized in that: The step of generating the first fixed-value matrix, the second fixed-value matrix, and the second distribution matrix based on the small-scale channel matrix samples includes: Based on the small-scale channel matrix sample, obtain the eigenbase matrix of the transmitter correlation matrix and the eigenbase matrix of the receiver correlation matrix; A third matrix is ​​obtained based on the eigenbase matrix of the transmitter correlation matrix, the eigenbase matrix of the receiver correlation matrix, and the small-scale channel matrix sample. Based on the expectation of the third matrix, obtain the first constant value matrix; The second constant value matrix is ​​obtained based on the standard deviation of the third matrix.

3. The method according to claim 2, characterized in that, The step of generating the first fixed-value matrix, the second fixed-value matrix, and the second distribution matrix based on the small-scale channel matrix samples includes: The distribution of the small-scale channel matrix samples is fitted to obtain the probability distribution model of the small-scale channel matrix samples; Based on the probability distribution model, the base and exponent of the power corresponding to each element in the second distribution matrix are determined; the base of the power follows a generalized gamma distribution.

4. The method according to claim 3, characterized in that: The probability distribution models include complex Gaussian distribution, complex Student's distribution, and complex Laplace distribution; The step of determining the base and exponent of the power corresponding to each element in the second distribution matrix according to the probability distribution model includes: When the probability distribution model is the complex Gaussian distribution, the exponent of the power is 0; When the probability distribution model is the student distribution, the exponent of the power is the first value, and the base of the power is obtained by fitting the distribution of the small-scale channel matrix sample. When the probability distribution model is the complex Laplace distribution, the exponent of the power is a second value, and the base of the power is determined according to the first parameter and the second parameter; the first parameter is obtained from the peak coefficient estimated by the discrete samples of the third matrix, and the second parameter is a preset shape parameter.

5. The method according to claim 4, characterized in that, After generating the first constant value matrix, the second constant value matrix, and the second distribution matrix based on the small-scale channel matrix samples, the process includes: Based on the first distribution matrix and the second distribution matrix, obtain a mixed distribution random sample; Based on the small-scale channel matrix sample, obtain the target probability distribution function; Based on the probability distribution function of the mixed-distribution random samples, obtain the fitted probability distribution function; Obtain the degree of difference between the target probability distribution function and the fitted probability distribution function; Based on the mixed-distribution random sample, the first parameter and the second parameter are adjusted according to the degree of difference to optimize the mixed-distribution random sample.

6. The method according to claim 5, characterized in that, The step of optimizing the mixed-distribution random sample by adjusting the first parameter and the second parameter according to the degree of difference includes: When the difference is greater than the threshold, the first parameter and the second parameter are adjusted within a preset range until the difference of the mixed distribution random sample corresponding to the adjusted first parameter and the second parameter meets the condition of not being greater than the threshold. For the difference degree that meets the conditions, obtain the mixed distribution random sample determined by the first parameter and the second parameter corresponding to the difference degree.

7. The method according to claim 5, characterized in that, The step of obtaining the difference between the target probability distribution function and the fitted probability distribution function includes: Based on the target probability distribution function and the fitted probability distribution function, obtain the forward KL divergence and the backward KL divergence; The difference is obtained by summing the forward KL divergence and the backward KL divergence.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images